Ryzen Thread - Faster than 7700K, Pair with 6900K, 1700x $389

Continuing from old thread, laptop guy who overclocks here. Decided to bench my processor and even with a cooler that hasn't been cleaned in 4 months I still beat the desktop i7-3370 in multi threaded performance.
http://valid.x86.fr/pidssq
You can see by the multiplier that it can on occasion reach the full 4Ghz boost clock. Also have it on balanced power mode. Didn't notice till I was about to post. Dang it.

>>58910155
*i7-3770 My bad.

File: 1469883273122.jpg (73KB, 960x960px) Image search: [Google]

73KB, 960x960px

First for, watercool Ryzen or GTFO.

AMD unlocked all chips, but you were still too lazy to overclock.
So guess what?
CPU overclocks for you!

That's right these chips automatically overclock. So they can attain speeds that would normally BSOD you for a time, PROVIDING you give it the ability to stay cool. Not 70c cool, but as near room temp for as long as possible. Overclocking scales with temperature.
Say at 20c it'll hit 4.3ghz. but as temps raises to 40c it'll go back to 4ghz and as it hits 60c back to stock speed.
An air cooler is only capable of removing so much heat at a time. But with water, it'll actually store that heat over a wide space allowing you to stay at a cooler temp for much much much longer giving you boost clocks off their na-na and stability at high and hot clocks manually set if it's a carryover function.

>The Precision Boost feature works alongside Pure Power to optimise performance and enact on-the-fly clock adjustments without queue drains which provide higher performance at the same power draw. Also, the CPU can be tuned with 25MHz increments. Next up is the Extended Frequency Range (XFR) which automatically scales the frequency based on the thermal loads and cooling apparatus while permitting frequencies beyond the standard Precision Boost limits.
http://www.eteknix.com/amd-ryzen-architecture-features-revealed/

Air coolers BTFO

>>58910091
who was on internet during athlon days? how was it? no leaks? did bulldozer leak like this?
I need some estimates from history.

its not like there are even any good games to play

pc gaming is dead

>$389 chip
>faster than $300 chip
Well it should be I've seen i7-7700k on sale for like $309 just last week.

File: 1325455860605.png (60KB, 627x620px) Image search: [Google]

60KB, 627x620px

>>58910244
>An air cooler is only capable of removing so much heat at a time. But with water, it'll actually store that heat over a wide space allowing you to stay at a cooler temp for much much much longer giving you boost clocks off their na-na and stability at high and hot clocks manually set if it's a carryover function.

So you have to break up your load intensive tasks so your water cooler can cool down again after it reaches the point where it's basically the same as an air cooler?

File: 1465786445037.png (1MB, 800x1024px) Image search: [Google]

1MB, 800x1024px

>>58910324
It gives a higher headroom for 'burst' clocks. Where as air has a tendency yo reach it's thermal intake point quickly.
For most users they'd only peak out CPU's for 1-5 mins at a time.
To put it in perspective. My 8350 was reaching 70-72c on air in around 10-20sec's. But on water it's at 38-39c and takes 2-5mins to get to that temp, because there's much more matter to soak up that heat. That time it takes to soak up heat is beneficial to the new XFR function. The quicker the chip heats up, the less time of boost you'll have.
And we all know lower temps are best for clock speeds.
Really this is a game changer. Literally just equipping a better cooler is an upgrage to the tech illiterate, as it will result in overclocking by default.
The real aim of the game will be keeping the die temp as low as possible for as long as possible. And Water is what gives you that advantage there.
The auto clocking will go by temperature overheads. As in how much can I clock at at this temperature in this instant. Not a predefined max that is hard set.
Air cooling blocks will not be able to 'respond' to burst frequency's like water cooling. As they have a heatsink that is much warmer than watercooling.

>>58910496
I think we'll have to wait and see. The majority of people using Ryzen probably won't be using liquid cooling. I doubt they'd design that feature to be dependent upon it. I think you'll see decent boost clocks on air.

>>58910496
Not other anon, but I would kek so hard if their drivers would work poorly and set the pc on fire (and the house) because of shitty drivers.

File: 1465625726115.jpg (20KB, 500x357px) Image search: [Google]

20KB, 500x357px

>>58910558
Yes, air will still see a free overclock curiosity of XFR. But water will get more advantage out of it for much longer.
We'll know more when time comes. But for now it's really fun speculating on this feature.
I already know testing in the AMD labs was done with 400W pelter plates. So, hint hint.

File: why not.jpg (78KB, 640x480px) Image search: [Google]

78KB, 640x480px

>>58910584
Good thing they're not made by nVida then. They'd be back to 1Ghz by the end of 2018.

>>58910244
You're retarded dude. Both have a limit to how much heat they can remove. The difference is that it takes longer to warm up the water than the metal, there are plenty of air coolers that can keep up to within a degree or two with your standard AIOs

>>58910244
The NH-D15 says hi :)

File: 85275.png (37KB, 650x550px) Image search: [Google]

37KB, 650x550px

>>58910664
Exactly.
Under an absurd 340w load an NH D15 will show less than a 30 degree temp delta over idle.
A 22c room and the CPU is only at 50.8c with a fucking 340w power draw.

Big tower coolers are more than enough.

>>58910244
>I have no idea how watercooling works

File: NH-D15.png (180KB, 557x454px) Image search: [Google]

180KB, 557x454px

>>58910714
The pricing tho, Ouch!

>>58910942
They're only $85. The pricing on Newegg is fucked up from some chink vendor.
https://www.amazon.com/Noctua-NH-D15-heatpipe-NF-A15-140mm/dp/B00L7UZMAK

File: 1479653606015.jpg (54KB, 482x714px) Image search: [Google]

54KB, 482x714px

>>58910664
>The difference is that it takes longer to warm up the water than the metal
That's exactly the point. Water will net you a longer boost time before warming up causing the boost scaling to go down. These chip's aren't a flat overclock. They're dynamic.
>>58910714
Yeah, for a non dynamic overclock. Where it'll operate at the same frequency no matter the temp. The point of XFR is to take advantage of low CPU load time where the chip and cooling system is colder and maximize performance should you need it in that instant. By having more sink you can prolong this boost.
If you're running at 100% all the time this effect will 'time out' as the system gets to operating temp. But for most users that's not the case.

With Ryzen it's not a matter of overclocking for 100% load sustained for an hour anymore. It's about tuning to your needs, and taking advantage of low temps if ya got em, or can get them, and sustain them.

I'm glad for AMD that they perform so well and they are so much competitive with those prices, but for people in general I don't think that's important. Most people have a CPU <$400, and most people I know have an i5 or similar.
I'll just wait to see AMD pricing on their CPU that will compete with the intel i5 and some benchmarks ofc

>>58911093
AMD will most likely be putting out 6 core products at prices similar to the Core i5 range.

>>58911075
What is wrong with the way your brain works?
The cooler can sustain a chip dumping 340 watts of heat into it. It can handle that over any time frame, whether its years or microseconds. You'll never run into a scenario where the cooler would fail unless you put more heat into it, which is pretty unrealistic.
The VRM on the board will fail before the CPU cooler.

>>58911093
I'm probably going to upgrade from my 2500k to a 1700x unless the 6 core has better performance per core.
As long as future benchmarks look good and I can use my NH-D15.

>>58911075
So if I'm doing some intensive task over a long period of time, such as playing a video game for several hours, or do a video encoding job set to take hours, the advantages of water cooling with XFR can basically be thrown out the window because the CPU begins to throttle itself anyway. Either that or it remains at max boost clock indefinitely anyway because the cooling solution inplace doesn't let the CPU reach any higher temps and we're hitting XFR's built-in ceiling. Is that right?

>>58910244
Closed loops suck big time dude.
Not because the radiator size, the tubing or the fans but because the fucking pump is inside the CPU block.

You can have water at 20ºC and the CPU will still hit highs 70ºC because the CPU block is too cramped and heat can't be transfered fast enough.

By the way, when you see a "liquid temperature" report from a closed loop what you are seeing is actually the temperature of the copper plate inside the CPU block and not the water itself.

Also If you think this auto OC is going to be better than a manual OC you're delusional.

To begin with this will be more conservative than a manual OC just by the simple fact that not all the silicon is the same, some will clock higher than others, some will have more voltage leakage and more heat output.

This XFR sounds like what motherboard manufactures have been doing with the UEFI auto OC utilities but now built in the CPU.

>>58910244
The entire point of water cooling is to increase radiator area. Heatpipes carry away heat better than water so unless you are using a large radiator water cooling is pointless.

File: 1482841710478.jpg (338KB, 1190x1340px) Image search: [Google]

338KB, 1190x1340px

>>58911107
Yeah, but it's at a disadvantage because it's still got a fair amount of warmth down at the head. Making it quicker to reach a higher temperature than water cooling that has water remove the excess heat as it passes by. Making the system temp raise as a whole and taking longer because of volume. Netting sustained lower temperature that will give you higher clocks for longer from the XFR function.
>>58911139
Yeah. As the die temp raises XFR will adjust clocks to suit. But you'll still get an advantage for the first 5-10 mins while the water heats up. But on air it'll get to it's upper operating limit a lot faster thus lowering the clock speed in the first 10secs-2mins depending on loads ect.

>>58911154
>This XFR sounds like what motherboard manufactures have been doing with the UEFI auto OC utilities but now built in the CPU.

Its not, but the other retard has no idea what it is either.
Its not auto adjusting base or boost state clock speeds. The CPU just has no fixed upper point in its pstates. If the chip is kept cool enough, and it can hit 4.3ghz for a micro second while staying inside of TDP, it will do it. The more cooling you have the more frequency you can gain, but this isn't magic. A transistor is still very much a physical thing with finite properties. Like anything else you reach a point of diminishing returns where increased cooling capacity stops netting you frequency gains because you've already eliminated a high degree of leakage in what is already a low leakage device.

Its just giving you average higher clock speeds under load when available without increasing power consumption. Cooler transistors will leak less and can switch faster, so XFR exploits that.


>>58911194
XFR does not increase TDP. A 95w chip with a 95w power target will never pump absurd amounts of heat into a cooler.
A Noctua NH D15 can handle a 340w load and laugh it off. A 95w chip is nothing for it.

You're just dumber than a box of Celerons.

>>58911104
>>58911129
I don't know if more than 4 cores matters that much regarding what I do on my computer. I think all will be a question of price and if there is similar performances on the programs I use daily.

How fucking disappointing. Guess I'll hold on to the 2500K for another 5 or so years.

File: 1481114468421.jpg (99KB, 810x800px) Image search: [Google]

99KB, 810x800px

>>58911212
>XFR does not increase TDP
I never fucking said that.
> A 95w chip with a 95w power target will never pump absurd amounts of heat into a cooler.
You don't say.
> The more cooling you have the more frequency you can gain
So then he was right al along.
>but the other retard has no idea what it is either.
You've literally only said what he has. Stop tugging yourself.
He's been explaining how to take advantage of the function by maintaining a lower temp for longer and pissing off air heads in the process. All you've done is confirmed everything he's said.

>>58911264
Have fun waiting for the new Intel architecture coming in 2021.

>>58911278
They come out with new sockets etc all the time. I'll be fine.

>>58910091
What baffles me is that despite being a AMD user for decades, they never actually cared for the ITX form factor.

I'm forced to used Intel for decent performance on ITX.

>>58911275
>responding to posts directed to two different posts as if they were all directed to one
Trying to defend yourself by pretending to be another person?
If this isn't a samefag fail its top tier retardation on display.

>He's been explaining how to take advantage of the function by maintaining a lower temp for longer
He(you) has been rambling like a schizophrenic, stating things that are flat out false about how a heat sink works. You don't need a huge water cooling loop which is his argument. Its not even remotely true.
Keeping at chip at 40c vs 45c isn't going to give you any real gains in sustained frequency. XFR is providing small boosts, not massive frequency gains. Leakage power as a percentage of total power is incredibly low for FinFET devices, keeping the chip in a normal operating range is going to give you the same benefit.

You must have gone to special ed on the front side bus.

>>58911298
What does a socket have to do with anything?

>>58911315
https://www.youtube.com/watch?v=y-bYSC6OT6s

>>58911320
As in
>they change sockets all of the time anyway so there's no optimal time to upgrade with Intel
and
>New Intel architectures haven't resulted in massive IPC improvements since Sandy Bridge

File: 1145930595777.jpg (212KB, 800x1181px) Image search: [Google]

212KB, 800x1181px

>>58911318
>XFR is providing small boosts, not massive frequency gains
>>58911212
>If the chip is kept cool enough, and it can hit 4.3ghz
>for a micro second while staying inside of TDP,
Because all we know thus far is that "Extended Frequency Range (XFR) which automatically scales the frequency based on the thermal loads and cooling apparatus while permitting frequencies beyond the standard Precision Boost limits"
>thermal loads and cooling apparatus
>and cooling apparatus
>cooling apparatus
Keeping the chip cool will always allow higher clocks while staying inside of TDP. The cooler you can keep the chip, the more hertz you can have per a given set of parameters. Now booting to your theoretical 4.3 will as a result in more heat raising temps that then in turn will throttle back XFR. But it's doubtful that it will crash back to the stock frequency when there is still thermal overhead to take advantage of.
>You don't need a huge water cooling loop which is his argument. Its not even remotely true.
Thermal dynamics. Hot into cold > hot into hot. That's where the base design of the air cooling heat sinks falls short in this system of a dynamic CPU.
This is new tech, it does not conform to traditional overclocking principles where you take it to the max and tweak out frequency there till it's just stable at the best temp the cooler you've got can dissipate. It's close to being the opposite. To take advantage you need temps to swing slowly. The slower it take to get to max temperature the better. But you cannot see this for some bizarre reason. Cognitive dissonance, look into it.

File: 1477362801322.gif (708KB, 951x534px) Image search: [Google]

708KB, 951x534px

>>58910244

>>58911458
Schizo:
Chips tend to run hot at high frequency not because of the frequency, but because of the supply voltage. Raising frequency alone has negligible effect on power consumption and heat output when supply voltage is fixed.
What keeping a chip cool does is lower leakage current so less voltage is wasted at a given input level. Less wasted heat will give marginally higher thermal margin to exploit.
Again, a D15 can handle an absurd 340w load while having a temp delta of less than 30c. 340w. We're talking about exploiting tiny momentary frequency uplifts in a 95w chip, which is still going to be limited by its power target. This is not a matter of hot into hot, the heat sink is fucking cool. 50c at 340w is cool. XFR is calling for a tcase max of 60c, and a D15 is keeping a 340w chip at 50c.
Pull your head out of your ass.

Inside of a 95w power target there will still be a practical max frequency that can be achieved, cooling the chip beyond a certain point does nothing. You are insisting that a water loop is necessary, and this is based on literally nothing but a product of your mental illness. You hit the power limits before you hit thermal limits, and frequency can't increase any more. XFR is not magic. Neither is water cooling. I doubt you've even taken a physics class in your life, let alone gotten a diploma of any sort.

>>58911458
Why you are defending a 5 minute performance boost so hard is beyond me. Why dont you fucking run a phase change cooler then, that'll actually have a real world,effect. Its a gimmick if the performance boost of a few percents is for 10 minutes.

Your thermal dynamics explanation is retarded too. Like seriously retarded. You neglect the fact that all water loops will reach an equilibrium temperature, where your hot > cold trip falls apart. What makes a difference is ducking surface area, do you think heat pipes are really that inefficient at transferring heat?

A 5-10 minute boost is a fucking gimmick.

The 1700 feels like the one to get, so long as its priced fairly. Hopefully it'd be a nice leap over my i3.

Not sure what GPU to pair it with however. Since i'm still slumming it at 1080 i'm pretty sure just a 480 will be fine rather than going "all in" with vega.

>>58911585
>A 5-10 minute boost is a fucking gimmick.
if it's activated on higher loads it's not. Doubt it does that.

>>58911605
$360

What the fuck is this watercool assburgers problem?
Spend shitload of money to theoretically gain a marginal performance gains after a COLD BOOT. Shits fucking retarded

File: 1484697378120.png (2MB, 1279x720px) Image search: [Google]

2MB, 1279x720px

>>58911585
>>58911611
>What the fuck is this watercool assburgers problem?
$80 nocturna VS a $100 AIO. Kinda self explanatory. Then theirs the want for maximum potential out of a new feature. And this is a board for the discussion of tech. I hope you had an interesting read and even learn something.
>Why you are defending a 5 minute performance boost so hard is beyond me.
On behalf of normalfag gamers that don't know how/are afraid of overclocking. They're the ones who will benefit from it.
>Why dont you fucking run a phase change cooler then
Already looking into sub cooling via peltier plates.

>>58911606
You're not going to get any additional frequency when you're loading all cores in Prime95. The model that justifies something like XFR existing is average frequency over time in selection of nominal workloads and games. If you get 3%-5% higher average clocks for no power consumption increase then why not? Thats what it does. Its not some dramatic game changing thing, but its worthwhile for a brand new arch trying to squeeze out ever ounce of performance to impress reviewers.

>>58911611
Maybe some serious post purchase rationalization, or maybe hes just nuts.

File: C4Xf4YgUcAALBE6.jpg (103KB, 1024x684px) Image search: [Google]

103KB, 1024x684px

>>58910091

I really hope we arnt raked by motherboards. Ideally the chipsets will be cheap enough that they can undercut Intel significantly.

>>58911632
No. Normalfags have nothing to gain. Nobody plays for 5 minutes. U're retarded

>>58911632
On behalf of normal fag gamers who probably play for extended periods of time.
Do you think a control scheme that over shoots what it can sustain is a good thing?

>>58911661
Top end X370 boards are allegedly in the $200 range, not priced like high end X99 boards.

File: 1479721785373.png (31KB, 420x420px) Image search: [Google]

31KB, 420x420px

>>58911668
I'm sure theirs enough games that max out CPU's infrequently enough to warrant it. And will only be more in future.
>>58911663
>All games run CPU's at 100% all of the time
Oh how on earth does someone argue with that infallible logic.

Does anyone know if the heatsinks are going to be compatible with AM3+ heatsinks?

>>58910256
I'm pretty sure I saw a couple of Bulldozer leaks back then. I don't think I saw any back in the Athlon 64 or 64 x2 days though.

>>58911700
>Oh how on earth does someone argue with that infallible logic.
You have no logic

>>58911642
it might increase average low in games, since it would only boost in heavy scenes

>>58911700
XD muh cpu isnt maxed out but I need more performance
If a game spikes and demands more cpu load you might be playing a shit game that was made by retarded code monkeys
Its a gimmick and why do you assume games will take less cpu in the future? Do you think cpu utilization will drop with new games on the same chip?

>>58911708
nobody freaking measured anything in CES, and they could
Everyone announced new brackets for am4, might be just marketing or they put something near socket that would block old am3 brackets.

In same boat, I really do not want to buy a new cooler.

>>58911708
>>58911752
AM4 socket is wider so AM3 backplates are incompatible and needs to be changed. HSFs with only clip mechanism are compatible

>>58911767
and when new proper am4 coolers come to my hellhole nobody knows
I'd be able to buy cpu day one but not cooler
box cooler it is

>>58911748
I meant games conforming to a thermal arc that isn't keeping it at high degrees all the time, so when it does kick up processing requirement XFR can come into play.
>XD muh cpu isnt maxed out but I need more performance
The fuck? I think you need reading comprehension classes.

>>58911779
AMD Wraith cooler isn't that bad imo. Its close to a evo 212.

File: makeitstopdog.jpg (46KB, 400x533px) Image search: [Google]

46KB, 400x533px

>>58911514

>>58911782
It's very good, especially compared to Intel's abominations.

>>58911811
Remember: Water cooling is inherently garbage that will inevitably break.

>>58911781
>scenes don't last for more than a few seconds
Do you really think that a cpu that's under moderate to high loads won't immediately rise in temperature when being pegged, and more frequency being pulled out of it? The rate at which a dye temperature will fluctuate is high

>>58911833
that's how GPUs work though, they seem not to care at all

File: AMD-Ryzen-Chart-6.png (13KB, 1214x745px) Image search: [Google]

13KB, 1214x745px

on single core it got same score as 6950k

amd on 3.4ghz v 3.0ghz intel. pretty disappointing desu

>>58911870
Synthetic benchmarks don't translate to real world performance, no game on the planet runs a the core frontend at 100%, you'll more than likely see that 12% difference in clock/perf drop to 2-4% in "games"

>>58911870
All core turbo on Broadwell-E is not the base frequency, and the Ryzen chip is not utilizing its turbo.

>>58911870
The Broadwell-E chip will be using turbo boost, probably to 4 GHz in that single-threaded benchmark.

File: Ryzen vs Kaby Lake.jpg (188KB, 600x728px) Image search: [Google]

188KB, 600x728px

>8 thread Ryzen costs as much as a i3-7350K
>12 thread Ryzen costs as much as a i5-7600K
>16 thread Ryzen costs as much as a i7-7700K

>>58911915
The 6950X can hit 4ghz on one core, its all core turbo is 3.4ghz.
The 6900k hits 3.7ghz on one core, 3.5ghz on all.
The 6850k hits 3.8ghz one core, I think its 3.7ghz all core, but I can't recall off the top of my head.

The scores there are all over the place, variance probably from different test systems. Thats the issue with using a shitty bench like Passmark. The results simply don't make any sense. In a single core workload the 6950X should always come out on top if the systems were equal.

File: 2017-02-11-12_30_57-PerformanceTest-9.0-Evaluation-Version.png (75KB, 1607x1391px) Image search: [Google]

75KB, 1607x1391px

>>58910091
>>58911870
Those benchmark numbers are fakes that someone generated and uploaded to passmark database. The idiot copied the cache numbers from wccftech article that has them listed wrong, the real cpus have 64kb L1I and 32kb L1D per core, not the other way around.

>>58911870
Graphic shows 6900k @ 3.20 GHz is faster than 6850k @ 3.60 GHz at single core mate.
It should be obvious that this is not an IPC benchmark but a mishmash of baselines with turboing going on at background >>58911946

File: 4.jpg (238KB, 720x540px) Image search: [Google]

238KB, 720x540px

>>58911964
Well you're right about the L1s being reported wrong there.

>>58911658
>ywn be so much of a shill that you marry your build

>>58910607
You know there is a parody point right?

let's say you have a d15, and you have 240rad aio

now lets prime95 this bitch

the d15 will hit its parody point faster, lets say its around 65-75c but the cpu just can not push it any higher than this. If you evacuate the air from the case, it will only intake cooler air, and you will never see a higher spike in use.

with water cooling you will at best extend the parody point out a few minutes where nothing you do can make that damn cooler any hotter

both can have that same parody point, and unless amd does something absoulutely fuckign retarded, like spike up to lets say 5ghz sees the temp gets to 60c then kills off its entire oc down to 3.4 then once its cool spikes to 5ghz again, its likely going to do this

startup, need to relearn what it can do
wait for a load and push it, to boost and see what that does
once boost hits the parody point, lets push it higher and step shit up in increments waiting for a parody point each time and deciding can the cpu take more, and once it finds its point it cant take more but its comfortable at this speed, that's what it will boost to, until the parody is broken by dust or total room temperature forcing it to go lower.

At least that's how I would do it and I consider myself retarded so amd has to do something at least as good as what I just said.

File: AMD-Ryzen-7-1700X-CPU-Single-Threaded[1].jpg (59KB, 1084x481px) Image search: [Google]

59KB, 1084x481px

>>58911870
aren't these cpus overclocked in this image? i've heard that this one is without overclocks

>>58911154
depending on how smart the xfr system is, it may get you close to peak oc, or even potentially better sustainable oc, if the cpu has as many checks as it is said to, that will be a fuck load of data, and more so then up the multiplier and run a test... did ti pass, up it again, it failed? up the voltage and try again.

maybe the cpu can figure out on its own when something is fucking up or when something can be pushed harder and can do it automatically.

I mean yea, took good to be true but mother fucker we should always dream big and push others to do better.

>>58911179
NO, you fucking idiot NO
god damnit, do you people just fucking ignore what is on cases? look at this
https://www.newegg.com/Product/Product.aspx?Item=9SIA6ZP3KE5690&cm_re=thermaltake_x9-_-11-133-275-_-Product

you see those rubber parts? you fee the fucking tubes through that and take the radiator out of the fucking case, he'll, put the radiator outside of the fucking room.

If anyone has a custom loop and doesn't take advantage of this, they are fucking retarded.

I mean for fuck sake, you have literally ONE good use for a custom loop and every fucking faggot refuses to put the rads and shit outside of the case.

>>58911826
If you do it right, it wont, outside of wear and tear and you can say that about any fan

someone try to persuade me from being hyped for ryzen

>>58911708
athlon 64 leaked 8 months in advance by canard pc.

bulldozer we had them telling us it was worse single core, but so much better multi core, and it was, what no one expected is it would be WORSE single core then the prior cpu, the phenom II

>>58912507
Your mom will recommend you get it because it's cheaper.

>>58911782
its close enough that the only reason you go with an evo is because you want to put a powerful fan on it, hell, it's better than the phenom II cooler, you know, the cooler that amd gave you that was able to overclock with and was designed for 125 watt cpus.

>>58912520
my mom only wants the best for her special boy

>>58912482

My buddy takes advantage of the fact that he literally lives in him basement and uses high pressure pumps to a 55 gallon drum of ice cold water, no radiator at all. How's that for a custom loop?

>>58911233
head room, NEVER and i mean that fucking NEVER overlook how much more headroom is really worth, we are looking at a 2600k and questioning if anything is even worth upgrading to STILL while people who have an i3 of that generation... the pain they feel is real.

>>58912551
is the drum outside or what? how does he keep it cool?

>>58910091
Can't calculate primes for shit

>>58912551
Had the exact same idea, not basement, but just getting a 10-20 gallon bucket, put some bleach into it and add ice periodically

I mean seriously, if you have the ability why not?

I made the argument with my little brother when he built his, why not pump the water from his computer out to a external rad, and put the rad into a bucket of ice water and have a shitty water pump inside to spin the water, that way he could do without fans and could just dump some ice into it. snf never have his room heat up... but no, he went with a sub 100ml reservoir and put everything inside his case and his room get sweltering even with the ac on full blast.

>>58912567
depending on drum size, just evaporation alone from it would likely be able to keep it cool enough I mean it can dissipate heat faster then you could heat the entire bucket up unless you are rendering 24/7

even then its sub 200 watts trying to heat up a 55 gallon drum. you really want to keep it cold, you could just pump the water out and add in nice ice cold tap water.

>>58912567

Have you ever been in a real basement? Shit tends to be cold all year round.

>>58911339
intel hasn't had a new core arch for over 20 years, what they are using a heavily built up pentium 3

>>58912624
yes a few times. but the ambient heat has to dissipate some way.

File: pnfx9f.png (34KB, 403x402px) Image search: [Google]

34KB, 403x402px

>>58912507
Because a laptop from 2014 still kicking. I really need an upgrade soon. All this overclocking is making me giddy. Lets hope desktop chips in laptops will come back for the Clevo or MSI line. http://valid.x86.fr/pnfx9f

Really too bad they don't make laptops with new graphics cards with older processors. "How much do we need to pay to ensure you use our newer inferior processors over this older beast"

>>58912654
>Really too bad they don't make laptops with new graphics cards with older processors. "How much do we need to pay to ensure you use our newer inferior processors over this older beast"
A processor is more than just its performance. It's a platform. It has utility. It has new features. Faster USB ports, AES, virtualization support, that sort of thing. People completely ignore this like it's just whatever because the performance is close. Make sure you inspect each new feature.

File: 1271536856622.jpg (58KB, 341x386px) Image search: [Google]

58KB, 341x386px

>>58912654
>70 points off an 7-4770K for multi threading
>with a i7-4810MQ
>clocking 3790.95 MHz
>97 °C / 207 °F
Nigga.

>>58912695
I won't say you're wrong but feature wise the haswell series has everything people want right now besides some premium features that most gamers don't utilize. On top of that most laptops using the newer processors don't even provide ports to access newer physical features. As for processor features, this may matter but all I seen from intel is a few extra instruction sets and a better iGPU... I mean no one except the poorest of gaming laptops rely on an iGPU.

>>58910318
why do people keep doing this? CPUs are not only about speeds, they have other features, too. AMD usually throws all the ISA extensions in their new CPUs, while intel, being the jews they are, charges you for additional specs (vt-d, avx2, etc)

Ryzen's lookin good, so many haters

>>58912559
how are sandy i3 these days?
worse than having one core in times of dualcore intels?

I did have athlon when dualcores became a norm, it wasn't fun.

>>58910263
No anon, PC gaming, all gaming is fine. You just got old and/or jaded.

its going to have shit single core as expected. DOA

nothing shills can do about it

>>58912817
It's not so simple. Things come and go. They don't always get better every year. There's a big problem with PC right now, specifically with Steam, where it's so easy to make something and the filter is so weak that a lot of games are horrid. Remember in the past when this crashed the market? At least you've probably heard of it.

When whether it's a good year or not is based on a few good games coming around, sometimes those games just don't come. Occasionally you'll get something truly interesting like Dark Souls but if you aren't into esports or certain kinds of experiences you might not have a good time for quite long stretches.

I would argue the indie/retro resurgence at least partially happened because the market was growing tired of its options. Some corporate types have grown very disconnected from their audiences. Someone with a $50,000 budget, if that, shouldn't be seriously competing with a $50,000,000 product.

>>58913066
current gaming intelligence is all about open world/survival/world shaping mechanics. story games became duller after 2001, when the graphics competition started. i hope the real idea people come back after we peaked with the technological stuff like graphics and mechanics

>>58910091
>Faster than 7700K
Only in stuff that can make use of all the cores.

Single thread and things like gaming the 7700k might still come out ahead.

I know you guys like to hate on gaming and shit, but who do you think is the target market for these cpus ?
For every person that wants to make some workstation or shit, there are a hundred gamers for which single thread performance will mean the most to.

Let us wait and see for a proper review of ryzen before reading too much into a single benchmark.

>>58913286
gaming hasn't been single thread for years.

File: 04509977.jpg (25KB, 211x193px) Image search: [Google]

25KB, 211x193px

>>58913308
>he plays open world AAA trash

File: 1485994187938.jpg (32KB, 395x401px) Image search: [Google]

32KB, 395x401px

>>58913327
>indie games won't run on a potato

>>58913327
Games that are old enough to be single-threaded don't benefit from anything stronger than a Pentium 4.

File: 80041619.gif (32KB, 250x250px) Image search: [Google]

32KB, 250x250px

>>58913360
Ok, try Starcraft 2 on a pentium 4.

P4 IPC is trash even compared to shit like bulldozer so it can't compete with any generation of intel i3-7 even if they have only one core enabled, in any game.

File: sd.webm (2MB, 600x336px) Image search: [Google]

2MB, 600x336px

>>58913327
of course I do

File: arnie look.gif (3MB, 353x209px) Image search: [Google]

3MB, 353x209px

>>58913395
>anything stronger than a Pentium 4
>try Starcraft 2 on a pentium 4

File: 81865.png (38KB, 650x450px) Image search: [Google]

38KB, 650x450px

>>58913308
and yet a modern 2c ( i3 7350K) beats out an 4c/8t 2600K in gaming while using less power

anandtech tested out broadwell-e with cpus with 10 cores, and the extra cores made no difference in gaming
http://www.anandtech.com/show/10337/the-intel-broadwell-e-review-core-i7-6950x-6900k-6850k-and-6800k-tested-up-to-10-cores/8\

You can make that claim all you want, but the benchmarks show otherwise.
Extra cores in games don't mean shit, IPC and higher clock speed do.

>>58913470
>"in gaming"
>implying
>shows one benchmark and a wildly unrelated link
gj mate

>>58912817
You just think PC gaming is good because you're a newfag who only knows the way of modern gaming.

>>58913485
>wildly unrelated link
You mean the link that shows that there is no performance boost from having more cores ?

File: 1486844576923[1].jpg (76KB, 680x510px) Image search: [Google]

76KB, 680x510px

>>58912654
Replying to my own post; had something possibly constructive to add.
I was thinking, This intel Processor is only 47W and is capable of sustained 3.6GHz 4c/8t load with 3.8Ghz 4c/8t short-term benchmarks. (Could go higher if I took it outside to cool it with 35 degree air.) And can even hit 4Ghz without too much of a fuss. Unstable slightly if left to sit at 4ghz however.

What about those AMD 65W desktop processors. 18W more than my particular Intel part. Will they be able to overclock like that? Lets hope so, because 4Ghz processors at 65W with an IPC similar to broadwell-e would not only be great but also a viable "gaming" laptop processor. If and only if manufactures pick it up. The APU's may be nice but nothing can really make up for the power of two discrete units in tandem. AMD gaming laptops might become a thing again.

In particular based on this screenshot the 65W 8c processor would make a very valuable competitor to those $3000 Xeon based workstation laptops. Especially since the last time a higher than 8 thread laptop was offered by Sager / Clevo / Eurocom. (2013) That laptop was a Xeon E5 based laptop. That particular laptop line supported several V2 E5's with the maximum being 12c/24t 2.7Ghz. The 8c/16t offering was 3.4Ghz.

Current workstation laptops have E3's which currently max at 4c/8t with a max offered TDP of 80W. This means they're not much better than the average high end mobile i7 part.

I'm a laptop guy, I like laptops and generally purchase something based on how long I can push it along, and how long it'll last in general. The road ahead is looking bright for the possibility of 8c/16t laptops becoming a thing and at the price point of your average Intel based gaming laptop currently. Discrete graphics choice is up to the consumer. That'll fluctuate the price a lot.

Gaming's not everything but damn, a laptop that can do both? Sign me up.

>>58913492
You quote the 7350K yet provide no proof for your statement. Also, none of the tested games is DX12/Vulkan.

>>58913526
http://www.anandtech.com/show/11083/the-intel-core-i3-7350k-60w-review/7

>Also, none of the tested games is DX12/Vulkan.
way to move the goal posts

Seems like no matter what evidence I provide you will always find something to nitpick about to claim that the benchmarks don't matter.

>>58913573
But senpai your benchmark clearly shows that there are gains to be had when you're not in the GPU limit.

>>58912726
Oh but you get intel management engine for free with every intel cpu!

Can amd even compete?

>>58910244
Welp Fuck. Now I have to spend $200+ to get a decent loop.

Will ryzen have aes ni? Or amd-v? Or >>58913694 management engine counterpart (why)?


I want naples now.

>>58913395
Yeah Ryzen is going to struggle so much with Starcraft 2....

>>58913512
>they really went with a 3,5,7 naming scheme

>>58911782
I'm just glad they implemented a stock cooler than is actually good enough for normies, it devalues low end aftermarket coolers and to some extent, high end.

>>58913827
So as to not confuse Intelbabbies.

>>58912741
a $65 g4560 easily beats a Sandy i3 and an i5 in certain applications. Intel finally dekiked the pentiums and made them hyperthreaded again, so the new ones are 2/4 instead of 2/2. Plus they benefit from 2400 DDR4

>>58913827
I'm with
>>58913847
On this one. They could have picked a better naming system but this works for Intel and if their products hold up and work well we may actually start seeing things R7-1700x or R3-1200 become a standard way to note a Ryzen part and future Zen parts.

>>58913847
It would have been less confusing if they numbered them after the amount of cores.

>>58913876
Though I just remembered they have the Radeon Cards that use the R# designation. I guess the -1700x would give it away that it isn't a graphics card but still may cause confusion.
>>58913890
I like this idea better really.

Will Ravenridge (Zen APU with onboard VEGA) be the GOAT HTPC platform?

>>58913924
Yes. I suspect it will at least rival current consoles' performance with more flexibility.

>>58913942
And by current I mean PS4 Pro / Scorpio.

>>58913755
>cheapest cpu block 20€
>50m of garden hose 20€
cut the hose in two and connect it like: tap->hose #1->cpu block->hose #2->out of the window

>>58913988
Genius.

>>58913988
believe it or not we did this in a factory with temporary hydraulic intercoolers.

>>58914005
>55 gal drum free
>cheapest cpu block 20€
>submerged aquarium pump 20€
>hosing 20€
>fittings 50€
150W heat load that OCd CPU could theoretically generate would heat up the water by 1°C in ~1.5h. 8h gaming session Δ5°C.

>>58914196
assuming 0 drum-ambient heat flux ofcourse

>>58912741
While I wont call them unusable, even with a sandy i5 you feel an upgrade would be useful, but if you just use it as a facebook machine then you may not feel anything is wrong.

>>58913225
we are already peaking with what we can do graphicly, short of pathtraceing lighting is getting better.

and ideas... you are looking at indie games or small teams for that as AAA are almost universally focus tested with children.

>>58913988
>water bill

It would cost about $200 per month 16 hours a day even here, too much.

>>58913286
in the case of single core games, sandybridge is good enough and if it isnt, the game is broken by design

in the case of games using dual or quad cores, let me tell you this, the moment something happens in the background, you want load balancing, the overhead you get from more cores than the application can utilize is not to be overlooked.

then you have more modern games that are using all 8 cores, or at least engines that can utilize them and with amd pushing real 8 cores to reasonable prices and likely getting some marketshare for it, game devs will take notice, either because 8 cores are i7 prices, or because 6 cores are i5.

>>58913360
doom with mods.

>>58913815
everything does, would it kill blizzard to make a good engine for once in their god damn lives?

>>58913470
thats not showing quality of life, its funny how games benchmark so high, but they never really take into account frame drops, which 6 or 8 cores smooth out completely.

>>58914016
Its a fantastic idea if you got the money for water/the need for it to be constantly cool

>>58913837
nah, the value of a 212 even if it is about equivalent in cooling is ease of cleaning, along with larger fan being quieter, or even just higher performing, and then a good dual tower aftermarket to keep shit cool even after it self oc's

all amd did was make the boxed cooler not a colossal piece of shit.

>>58913985
probably not, scorpio will have the equivalent of a 480 in it. as much as I would love to have that on an apu, I just don't think they would do that.

>>58910244
Does that mean that I need to buy five 520mm radiators, 40 140mm 3000rpm Yate Loons, and three pumps for my Ryzen cooling loop?

>tfw AMD is going to back to it's Athlon 64 days
'm going to keep my 4690k but i'm so fucking glad AMD is upping their game

Gonna start a review channel for fun when Ryzen comes out

What are some comparisons /g/ would see rather than the usual gaymer /v/-manchildren shit everyone else is pushing?

>>58914619
Undervolting, Zen vs Kaby/Skylake at sub 3.0GHz clocks, primarily interested in power consumption, performance is nice too.
Single thread tests with disabled cores to see how larger L3 caches on low core counts affect performance.
Memory benchmarks, effect on performance of high performance DIMMs, bandwidth vs latency.


Primarily that, everything else gets tested by review sites.

File: 6f6.jpg (125KB, 1024x576px) Image search: [Google]

125KB, 1024x576px

>>58914259
>focus testing with children doesn't result in innovation
I would argue a lot of the best ideas could use the creativity and open thinking of children. If anything, it seems like the games that take themselves too seriously are the ones that come out stale more often than the other way around.

I agree with the bit about small teams and indie though, despite most of them being shit.

File: Hardcore Torture Porn.webm (349KB, 1432x1080px) Image search: [Google]

349KB, 1432x1080px

>>58911811

>>58914676
>Zen vs Kaby/Skylake at sub 3.0GHz clocks
That makes no sense. Clock speeds usually scale pretty well for most application, so if Kaby Lake does .05% better than Ryzen at 4.0GHz, the same should be seen at 2.0GHz.
Although power consumption is a good idea. I think I'll wait to see if there is an AMD equivalent to the "T" variants with Intel's line-ups. I have an i3-6300T.

>Single thread tests with disabled cores to see how larger L3 caches on low core counts affect performance.
Biggie
>Memory benchmarks, effect on performance of high performance DIMMs, bandwidth vs latency.
Also a biggie, but again, performance with memory frequency also scales pretty linearly for the most part.

>>58914775

>That makes no sense. Clock speeds usually scale pretty well for most application, so if Kaby Lake does .05% better than Ryzen at 4.0GHz, the same should be seen at 2.0GHz.
>Although power consumption is a good idea. I think I'll wait to see if there is an AMD equivalent to the "T" variants with Intel's line-ups. I have an i3-6300T.

That's why I said I'm primarily interested in power consumption to know the perf/watt sweet spot of the cores and to see which one had to 'overclock' from their initial design to attain a performance target

>>58914809
I'd honestly wait for the low-power versions of Ryzen before doing that test. Usually, undervolting and underclocking leads to some weird power consumption issues on Intel's CPU (I tried undervolting and underclocking a Pentium G3258 and it did some funky things with Turbo mode deactivated).

Also, would it be fairer to compare the R7-1800X with any LGA 2011-v3 CPUs over LGA 1151s? That also makes a big difference with power consumption since the iGPU on LGA1151 CPUs always sip some amount of power.

>>58914809
And also remember that Ryzen and Kaby Lake have different design requirements which makes this sort of comparison fruitless since we're comparing apples to oranges.

File: 15_20.jpg (674KB, 1600x1336px) Image search: [Google]

674KB, 1600x1336px

mfw Intel goes into full crisis mode when 1700X and 1800X is competing against their top i7 models at half the price.

mfw all new gaming machines will have Ryzen cpu's and soon to be Vega GPU's.

At the moment I run a 5820K and 980Ti, but the Intel and Nvidia prices are getting more and more ridiculous every year and I'm fucking glad AMD is actually going to make something awesome again.

This will all benefit the consumer. Removing the Intel/Nvidia monopoly.

>>58914871
>(I tried undervolting and underclocking a Pentium G3258 and it did some funky things with Turbo mode deactivated).
I'm interested, got any details? Normally it should go funky WITH turbo, not without it.


>Also, would it be fairer to compare the R7-1800X with any LGA 2011-v3 CPUs over LGA 1151s? That also makes a big difference with power consumption since the iGPU on LGA1151 CPUs always sip some amount of power.
Whatever you have on hand, really, but make it as fair as possible.
Optimal would be vs the newest architecture(so 1151) since that's what most people will be buying.
As for the iGPU. it doesn't sip any power since it's all power gates if it's unused, and has been for years on all platforms.
What will actually skew the results is that Summit Ridge needs a dedicated GPU that will most certainly vampire more power than any iGPU and fuck with your stats, unless you're directly measuring from the power rails.

>>58914911
not just consumers, but OEMs will jump their asses right onto AMD's dick for this good of performance at these prices.

>>58914921
My Pentium would lock at the maximum frequency I set it to at idle, but would fluctuate like crazy if I started a program. So it draws more power at idle from the fact that it stays at its maximum clock. With Turbo mode on, it goes from the maximum clock to the Boost clock like crazy at idle. This was on a Gigabyte Z97 motherboard.

>it doesn't sip any power since it's all power gates if it's unused, and has been for years on all platforms.
Huh, maybe I was thinking of APUs in that regard.
>What will actually skew the results is that Summit Ridge needs a dedicated GPU that will most certainly vampire more power than any iGPU and fuck with your stats, unless you're directly measuring from the power rails.
What's stopping me from running ESXi and Horizon on both and remoting into the virtual desktops without using graphics cards at all?

>>58914990
>Huh, maybe I was thinking of APUs in that regard.
They don't use much power either, maybe the Trinity/Llano ones did.


>What's stopping me from running ESXi and Horizon on both and remoting into the virtual desktops without using graphics cards at all?
Nothing really, I just didn't consider that option at that time.

>>58915015
What sort of background music should I play over the video? Fuck rambling like an autist.

>>58915033
Anime OSTs

>>58915055
Thought so

>>58915033
Zweihänder.

>>58915055
Would Hentai opening and closing music be too much?

I really hope Ryzen OCs well, I've got a fat 360mm rad just waiting for one.

>>58915307
Why not.

>>58915339
Give me a good one then. Haven't watched Hentai since I discovered the glory that is God

>>58915354
Futabu

>>58915360
>Futabu
Jizzus Christ
https://www.youtube.com/watch?v=_k0Jc135ErI

>>58915389
>Jizzus Christ
Exactly.

>>58915406
I'm not even going to credit the music. I'll get rid of the vocals and leave the background track running. Only the chosen will know.

>>58915417
Better disable comments because if someone asks for the BGM name some faggot from /g/ will snitch.

>>58913924
Unfortunately HTPCs are deader than dead. Nobody is making decent HTPC cases any more. Best option is to have a big fat NAS with all your media on it anyway.

File: 1402065566623.jpg (43KB, 450x450px) Image search: [Google]

43KB, 450x450px

This looks very, very promising.

The only thing I still want to know is if these things overclock well.

If I can get the 6 core 12 threads up to at least 4.3 ghz, it will be fucking awesome.

If not, I'd rather get a 7700k and put it at 5 ghz.

Also, does dual channel vs quad channel memory make a big difference?

>>58915461

>Also, does dual channel vs quad channel memory make a big difference?

For the workloads where it does you'd be using a xeon anyway for that delicious ECC.

>>58914688
no, thats not what i'm talking about, you are thinking nintendo coming up with a fun spin on the fps genre,

what im talking about is
https://www.youtube.com/watch?v=BYowhad_1mw
becoming
https://www.youtube.com/watch?v=p4brPYMBNa0

because ea focus tested a M rated game with 12 year olds who said overstrike felt like something their younger siblings would play.

general rule of thumb, you focus test something, you get fucking shit in return. Oh no, this R rated movie focus tested bad with children, let's make it more appealing to them.

So fucking sick of focus testing ruining shit I was interested in.

>>58910607
>I already know testing in the AMD labs was done with 400W pelter plates
Don't they stated on one of official XFR slides about liquid nitrogen?

>>58913787
>amd-v
I thought every amd cpu nowdays get this

File: 1486333454659.png (770KB, 716x1135px) Image search: [Google]

770KB, 716x1135px

I wonder how software RAID would work between the two SATA ports in the CPU and the two extra on the chipset...

>tfw no quad channel DDR4 so APUs won't BTFO the Intel IGPUs

>>58915735
Hint: there's a memory tech AMD's been developing for quite some time.

>>58915711
SATA ports are never in the CPU, they're all from the southbridge.

>>58915755
HBM2 is expensive as all fuck, it would ruin the point of APUs which is to provide good graphics at a low cost.

>>58914619
I got a bench for ya, stock doom on one of the loaders, I just want to see the fps that it can push that game to, I think my phenom II stock pushes doom passed 700fps, and I think a few processors push it passed 2000.

something stupid that likely only I'm interested in.

>>58915767
>HBM2 is expensive as all fuck
SSDs were also expensive as all fuck in the beginning.

>>58915759
>Incorporating much of the southbridge into the SoC, the Zen CPU will include SATA, USB, and PCI Express NVMe links

>>58915067
love slave
https://www.youtube.com/watch?v=Ft98ySpDezU

Its a staple of the most annoying songs you can subject someone to.

>>58915785
>Implying
They still are


for me anyway

>>58910244
I dont think you now how cpu boost works

>>58915807
Well they got much cheaper.

>>58915788
Yes.
Soutbridge = I/O
SATA = non performance critical I/O
Ergo SATA goes on southbridge together with USB.
PCIe, NVMe are latency sensitive and go on close to the CPU cores directly.
Unless AMD got rid of that differentiation and put everything on the southbridge if the latency is low enough.

>>58915823
Okay fine fair point on both accounts but that would also imply HBM2 APUs would take a while to be worthwhile like say 2018-2020

>>58915850
Nah, earlier. Mass production is what matters. But actually, we should wait for Vega and see for ourselves.

How do you faggots think raven ridge will affect the laptop market? I hope it's fucking awesome.

>>58915869
Only if they fit in 4 cores and a semi decent GPU under 17W

Which likely isn't happening, but I can expect two highly clocked cores with a kickass GPU with actual 3D drivers.

>>58915869
Absolute murderfuck. At least during K8 times Intel's mobile platform was not shitburst-based. Not they have nothing new but shitty die shrinks and even then they have huge troubles with 10nm.

>>58914911
amd can't compete gpu side thill they mcm gpus, likely navi.
See, even if amd comes out with a titan xp killer for under 500$, everyone would just buy nvidia.

But you mcm the gpu, you now have the same capability as they do for the cpu end to make a fucking amazing gpu.

Personally with amd, All I would be focusing on is hurting my opposition unless I had something that clearly crushes them.

See nvidia pull fuck huge profits, let's put out a gpu that surpasses it for cheap and just erode their margins. just hurt them, that is all I would focus on for the time being till you either got market share back, take away that deep dependency on proprietary software, let them make it open, and fight them on an equal playing field.

>>58915869
You'll need to outbribe Intel first

And make sure OEMs don't gimp it with single channel memory like Carrizo

>>58915767
a single 1GB stack would provide almost double the bandwidth of DDR4 and not be very expensive.

>>58914358
>thats not showing quality of life, its funny how games benchmark so high, but they never really take into account frame drops, which 6 or 8 cores smooth out completely.
so any benchmark that intel wins at, it is not really winning since it doesnt show frame drops..
how much further are we going to move the goal posts ?

Why not just admit that contrary to evidence you *believe* that moar cores = better for everything

>>58915890
>See, even if amd comes out with a titan xp killer for under 500$, everyone would just buy nvidia.
No, if said titan killer would actually cost 500$ you'd never buy it because it'll always be out of stock.

File: am4-procs-100702362-orig.jpg (63KB, 1138x637px) Image search: [Google]

63KB, 1138x637px

>>58915837
>In the case of Summit Ridge, AMD will integrate the CPU with a memory controller (although the number of channels is undisclosed), as well as PCIe, SATA, and USB functionality.
It's like you don't even read

>>58915902
The thing is that actually integrating it and the interposer are expensive.
The HBM2 module alone probably cost like 3 GDDR5 chips.
Which isn't all that much.

>>58915767
what price is it? seriously we keep saying its expensive but compared to what? how much does it cost?

>>58915869
Laptop fag here, see my previous posts.
>>58912654
>>58913512
Chances for high end look very good. Low end will still be terrible I think.

>>58915600
ecc?

>>58915976
ECC, Error Correction Code

>>58915976
Error Correcting Code

Basically makes sure RAM doesn't develop cancer (silent data corruption)

>>58915976
>>58915987
I should have noted that ECC Ram can correct single bit errors and detect two bit errors.
Also this >>58915999
Checked

>>58915927
I don't follow, the southbridge is in the SoC, meaning it already does the function of lowering latency.
The only thing you can talk about is where some of the I/O is, near the cores or on the southbridge, it shouldn't matter since the SB is integrated.

>>58915916
the 4850 and 4870 and 5000 line all say otherwise, next to fucking no one bought amd they all went nvidia.

our best bet for amd is they get good professional use out of the gpus, and they continue to advance with that in mind, and sell us the scraps, like they did with polaris and are going to do with vega, this way the scraps can also be very cheap scraps to boot, at least till the mcm gpu happens, then amd could shit on even the most powerful nvidia with one sku, and due to only needing to develop one sweet spot sku, iteration could happen at a far more rapid pace, think the 7000 lineup, then the 290+, then the 285, and then fiji, then polaris and then vega, but each one of them was an entire lineup instead of just a small sku line.

>>58916045
>the 4850 and 4870 and 5000 line all say otherwise, next to fucking no one bought amd they all went nvidia.
Because back then marketing actually worked. Now you just need to pay a bunch of youtubers and boohoo, you're ready to sell your shit.

>>58916045
Wondering if we'll get tandem VEGA GPU's on a single board this time around or if VEGA will primarily be a single GPU set up intended for one GPU at a time.

>>58916036
I meant that there are SATA controllers within the Ryzen SoC, not that the SATA ports are directly linked to the CPU. I was wondering if there would be any latency or bandwidth issues caused by the split in SB functions between Ryzen and the X370 chipset when using RAID0.

As far as I know, Intel's RAID controller cuts off at around 1800MB/s. I doubt Ryzen will be able to exceed that either, or even get close.

>>58915908
go look up real world use of a 4 core i7 and a 8 core i7.

If I remember right the people who did this showed off dying light and farcry 4, with farcry there was no gain in fps but the dips and stutter were gone.

quality of life IS a very important aspect that does not show in a normal benchmark scenario, only in real use ones.

>>58916056
marketing worked then, marketing works now, and amd doesn't have the presense to change that.

>>58916119
No it does not, otherwise AMD would never gain their 30% GPU market share back. Even normiest of normies watch fucking tech channels nowadays.

>>58916076
if you want to make those claims then you should link to those benchmarks.


http://www.anandtech.com/show/8426/the-intel-haswell-e-cpu-review-core-i7-5960x-i7-5930k-i7-5820k-tested/6

Anandtech does the 8 core 5960x and while it does come out ahead in certain benchmarks (transcoding, rendering and a few other shits) it doesn't justify the pricetag for the marginal performance gains in gaming.

> that does not show in a normal benchmark scenario,
of course, of course, the jews are keeping AMD down again by not showing these super important benchmark
ignore all other benchmarks, no matter how numerous they are or their sources they don't mean shit

>>58915887
I would be happy with a cheap two core solution, maybe even hyperthreaded.

>>58915888
Aren't the low-end core m3 bretty gud or are they a meme? I know the Z series is pretty trash and weak as fuck but some of the celerons and pentiums look 'okay' but haven't had the pleasure to test them properly.

>>58915897
Yeah Intel's grip on the OEM market is pretty fucking tight but there are still AMD laptops, although not many.

>>58916294
>Aren't the low-end core m3 bretty gud or are they a meme? I know the Z series is pretty trash and weak as fuck but some of the celerons and pentiums look 'okay' but haven't had the pleasure to test them properly.
Intel's low-power mobile CPUs are bretty good, BUT, we know nothing about actual energy efficiency of Raven Ridge series APUs.

>>58916294
>Yeah Intel's grip on the OEM market is pretty fucking tight but there are still AMD laptops, although not many.
It also has to do with volume, intel can supply the demand from OEMs
AMD has to share GF capacity with all their other clients, like qualcomm

>>58916333
I'd be happy with their current performance if they were cheaper. Either raven ridge is as good and cheaper or they force Intel to lower prices, which I know hardly ever happens but I'd be happy with either.

>>58916349
Ah I didn't think of that. Does AMD usually have supply problems? Will the fact that it's a mature node help with volume?

>>58916373
>force Intel to lower prices
REMEMBER, INTEL NEVER LOWERS PRICES. NEVER. THEY WILL NEVER DO THAT.

>>58916386
They did like some 20 years ago.

>>58916349
Global Foundries isn't running 14nm parts for Qualcomm. Samsung is. Samsung also coincidentally bought all the first lots of their S835s. Purely coincidental.

>>58916373
>Does AMD usually have supply problems?
They haven't ran into this issue before. Fab8 can shit out 60k 14nm wafers per month, and they just made an announcement stating they're increasing capacity by around 20%.

File: 1457358905476.jpg (1003KB, 4500x4334px) Image search: [Google]

1003KB, 4500x4334px

>>58916386
YOU CANT SROP MY DREAMS FAGGOT

File: 1455122603501.jpg (32KB, 375x344px) Image search: [Google]

32KB, 375x344px

>>58916386
Well, let them get finished and bankrupt at their peril.

The one thing that concerns me is the fact that AM4 is still behind the times when it comes to overall connectivity.

>>58916432
Hotaru is cute.

>>58916457
It only doesn't support Thunderbolt and Optane, and I don't expect it to unless motherboard vendors put it in themself in the case of TB.

If you mean lack of I/O, remember that extra I/O can be put on the motherboard in place of the missing southbridge.

>>58915807
They are okayish but still not enough to replace HDDs. 120 bucks for 512GB is cheap enough for normies.

>>58916373
>Ah I didn't think of that. Does AMD usually have supply problems? Will the fact that it's a mature node help with volume?
I can't speak about their cpu/apu's since those have not been in demand, but their RX gpus were in short supply for a long time after launch.

On the consumer side intel has shortages as well, like the 6700k when it launched was priced above MSRP for a while.
Intel (and AMD) might have focused on business customers first and consumers afterwards.

Keep in mind GF's entire capacity is only slighlty ahead of intels.
Intel has 6 fabs that do 14nm, GF only has 1 at the moment.

>>58916598
>like the 6700k when it launched was priced above MSRP for a while.
Because Intel's 14nm had shitty yields for a looooooong time.

>>58910091
Why did you leave out physics anon? They're good cpus you don't need to cheat. Just show they're not going to be gaymen cpus.

That's fucking expensive for amdshit. Why would I buy that over an i5 for half the price.

>>58916781
>Why did you leave out physics anon?
Because 17-17-17-39 memory is shitty for Physics benchmark?

>>58916485
No, I meant that the buswidth between the CPU and the SB on AM4 is quite narrow compared to DMI3.0 on the LGA1151 socket. It's a severe drawback.

>>58915927
Will AMD have these features in their x300 line?

>>58916802
It impacts the performance of the Prime Numbers metric, and the aggregate score as well. Those three metrics are the only ones where the system is notably behind the competition.

>>58916802
Why would anyone test with those fucking bargain bin beyond poorfag DIMMs, with a A320 motherboard to boot.

What in the fuck.

>>58916835
Either to deflate hype(while it is fun, too much hype is dangerous ), or make someone look bad.

>>58916832
Yes, it won the other five.
>>58916835
Because it's the best they could get without getting a lawsuit or two.

File: am4-chipset-100702364-orig.jpg (86KB, 1133x625px) Image search: [Google]

86KB, 1133x625px

>>58916824
That's independent of the chipset.

I seriously hope that these numbers are real.

I'm on a i7-5820k and that's practically the only high end model that's in a sensible price range.
Next model is nearly 200% more expensive and the performance increase is like 20% or so.
If AMD can offer the performance of a 1000$ CPU for ~400$, this is going to shake the market pretty radically.
Hell, it's not just going to shake it, it's going to fucking reshape the market.
Many people are at a point, where upgrading their old 2500/2600k makes sense if this kind of a performance increase is offered at a reasonable price.

>>58916866
>CrossFIre
>ITX
How the hell would this remotely work?

File: AMD-Radeon-R9-Fury-X2-Dual-Fiji-GPU-Graphics-Card.jpg (81KB, 1024x671px) Image search: [Google]

81KB, 1024x671px

>>58916924
What? You thought dual GPU cards actually combined the GPUs together?

>>58916924
Single slot or double slot half width cards, but then it's already stretching the limits of ITX


tl;dr xFIRE/SLI is cancer no matter what

>>58916944
wat?

>>58916266
do you even know what a benchmark scenario is you fucking retard?

>>58916906
>Many people are at a point, where upgrading their old 2500/2600k makes sense if this kind of a performance increase is offered at a reasonable price.
That's me. My 2500k running at 4.6GHz has served me well for over 5 years now but I'm really itching to upgrade.

>>58916969
All dual GPU cards run on SLI/XFire in order to work. That tiny silver square near the PCIe connector is a PLX chip, which acts as a switch for PCIe lanes to go from one GPU to the other, or both at the same time.

>>58916150
they didn't, they sold a fuck ton of gpus but it sure as fuck was not to consumers.

>>58916990
I see, is the Titan X designed this way?

>>58917050
If memory serves no, but I can't be bothered to fire up a search engine right now.

I'm really happy with my 6700k. Will 4 years down the road be time to upgrade?

>>58912494
It will still break, and when it does it will take your entire system with it.
Meanwhile I have fans that have been spinning nearly 24/7 for 10y and still work

>the only way Ryzen won't be subpar CPU is if you spend at least another $200 on a decent cooling solution.
So at the end of the day Intel is cheaper? Huh, didn't see that one coming

>>58917199
Or just overclock the $300 8 core.
XFR is mostly for normalfags and laptops.

>>58917211
>laptops
No, not really.

>>58911129

You can contact Noctua if it doesn't fit on your new CPU and they'll send you all you need to mount it for free.

>>58917211
I still need an expensive cooling solution to get something out of it.

>>58917232
Much less than your 140W Intel.

>>58917050
No, only the Titan Z was built like this
That and the 690 and 590

>>58917243
I use the stock intel cooler on my non k processor and the thing never went above 65C

>>58917227
Yes, really, it's pretty much meant for burst loads, and that's what low power laptops CPUs mostly are, go look at base and boost clocks on Intel/AMD ULVs, some have a turbo 1.5GHz higher than the base clock, XFR can raise that even higher.

>>58917259
Good thing the stock AMD cooler is eons better than the stock Intel cooler so you're set to go.

>>58917271
Ryzen isn't going in any laptops. Fucking children, I swear.

>>58917308
>Ryzen isn't going in any laptops.
What is
Raven
Ridge?

>>58917291
No, I'm not set to go because Ryzen is not worth buying unless you're willing to invest on a custom cooling.
Might as well go with a 7600 non k or wait for a 8600 non k

>>58917259
>I use the stock intel cooler
Lmao I have some tin heatpipes cooler I bought for $15 and it keeps my 3570k under 55C under load.

Just get any cheapo cooler you poor fuck, your system and ears will thank you

>>58917141
If 4 cores last it should be good for a while. Maybe longer.

>>58917308
There are laptops that don't need an iGPU, they're 15" and bigger in most cases, a large market.
So yeah, if AMD wants it I don't see why they can't shove some 30W quad core in laptops to go with some discrete GPU.

>>58917319
Ryzen is a product line, retard. Zen is the core architecture.
Ryzen is not Raven Ridge.

>>58917341
>non k cpu
>custom cooler
No thank you, burning my money will bring me more satisfaction

>>58917360
But Raven Ridge may still carry Ryzen brand name.

>>58917362
So both poor and stupid.
Got it.

>>58917374
CPUs and APUs have totally distinct branding.
Quit trying.

>>58917375
There is LITERALLY no reason to spend money on a better cooler though. You can throw ad hominems all you like but I'm still right

>>58917404
Noise and temps.
Of course with you being deaf and stupid these really wouldn't matter.

>>58914196

> I didn't already think of that

>>58917431
>noise
I use headphones so I couldn't give a fuck about it
>temp
There is absolutely no point in better temps, I could fucking use liquid ice to cool it to -10C at load and it would still run at the exact same speed

>>58916419
even then, if GF cant match demand, it will run off to samsung fabs for the 14 nm.

>mfw Ryzen CPUs have adjustable TDPs up to 200W and XFR can boost two cores to 5GHz with that.

If fucking only
IF FUCKING ONLY

>>58917498
This would be a Intel killer feature.

>need serial performance
Done
>need throughput
Done

Which is why I don't think we'll see something like this.

>>58917498

The Stilt is such a cocktease.

>>58916532
not for mass storage sure, but its hard to beat 8tb for barely more then 200$

>>58917498
I'd march down to my local dealer and threaten him to put me first on the preorder list for the $500 8 core if this was true.
No fucking joke, I wouldn't even wait for reviews if I knew this is possible.

are we going to see ryzen laptops?

>>58917308
Please look at
>>58912654
and tell me high performance chips don't make their way into high end laptops. If the performance is there they will be created to suit that particular market.
>>58913512
This is me looking forward to the possible re-introduction of AMD desktop parts finding their way into the desktop laptop (Or gaming laptops even) market again.

>>58917498
That's how XFR is supposed to werk, but you need a proper cooling solution for that.

>>58917552
Only in 2kg+ laptops if AMD bothers binning laptop chips.
Which it probably won't if demand for desktop chips is too much.
I guess they see a bigger opportunity with Raven Ridge, we don't really know what kind of iGPU will be in there, how much shaders, how much flops?

>>58917552
Raven Ridge faggot

>>58917151
when an aio breaks, its either the pump or the fans so long as you weren't retarded with the hose.

congrats on luck with the fans, haven't had one last more than 5 years for me though, but then again, 24/7 100% speed kinds of does that.

>>58917572
The thing is I don't know if XFR strictly sticks to TDP limits and if AMD has the ability to unlock the TDP that much.

>>58911315
Motherboard manufacturers determine what boards get made, there hasn't been much demand for AMD processors in several years and ITX designs are harder and require more R&D so it's not worth the investment for a niche form factor for an unpopular processor.

>>58917498
This is important, but does Zen have per-core voltage control? Because I doubt it can do that without that feature.

>>58917656
>does Zen have per-core voltage control?
Y-yup.

>>58916924
APU+discrete GPU

>>58917617

The Stilt hinted at being able to basically have no TDP limits on the top end boards. Note I said hinted - he admitted to nothing.

>>58917638
It's not just that, it's that AM3+ was a difficult socket to shrink down into an ITX form factor. There were two massive chipsets to fit on the board, and both of which needed a decent amount of cooling and power delivery in order to work. Then there's the question of the CPU and other power components, all of which needed a hefty amount of real estate compared to Intel's sockets.

There were quite a few FM1/FM2/FM2+ and AM1 ITX motherboards though.

>>58917656
Almost everything in a Zen CCX has its own voltage plane, from individual cores to the southbridge, to each level of cache even the GMI links for infinity fabric.

>>58917498
I'm pretty optimistic this will happen. I expect some sort of Wattman-esque tool to set temperature/TDP limits for the XFR, and disabling cores/threads should be pretty simple.

>>58917713
He can't admit because he'd get sued.

Lets look at this from AMD perspective.
Such a feature is good.
Such a feature would give a VERY big advantage over Intel.
Such a feature would solve their ST woes and then some.

Realistically it makes a lot of sense to have this feature.
Now the only questions are: Was AMD competent enough to get it to work.
Does AMD have enough working dies to sell those in any decent volume? Because this feature depends harder on silicon lottery than anything else.

>>58917727
>Almost everything in a Zen CCX has its own voltage plane, from individual cores to the southbridge, to each level of cache even the GMI links for infinity fabric.
Fuck they sure try to fuck over Intel in every possible way. What's Infinity Fabric?

>>58917793
The improved HyperTransport

>>58917811
Is it better than Intel's bus?

>>58917793
Interconnect for accelerators, multi-sockets, FPGAs, etcetc

>>58917822
No one knows for sure, yet. All we know is that it has improved latency over the FM2+/AM3+ HT

>>58917822
Intel's Omnipath is also a pretty new interconnect architecture, so we don't really know.
And in markets where you're juggling 40000 CPUs with 20+ cores each, what makes of breaks you is the interconnect, the core comes second.

>>58917617
You can freely adjust power target on Ryzen exactly like you can with Polaris based GPUs, and the chip will find the highest possible frequency to fit the target. XFR gives tiny frequency boosts above the top pstate when it thinks it can fit them in. It will always stay within power target. If it didn't it would just be a dumb overclocking feature. No one cares if a chip increases clocks while also increasing power. XFR is about increasing clocks without increasing power, but it'll do it at any frequency range so long as the chip isn't near its tcase max.

>>58917790
TheStilt is a literal nobody

>>58917985
>TheStilt is a literal nobody
He's a nobody with a sample.

>>58918001

He is also someone who sits in the top of lots of overclocking charts. That said he did spend a lot of time last year claiming ryzen couldn't clock over 3ghz much like juangra claimed.

NEW THREAD

>>58918041

>>58917985
Even at 95W limit I don't see how XFR can't boost a single core far over the turbo clock(lets say 300MHz) if all other cores are underutilized and are drawing symbolic amounts of power.

>tfw AMD had got your back.
>tfw Intel shills are being btfo
>tfw for the next ten years we will be living at the mercy of AMD

>>58912549

The answer is still AMD

>>58913897
>>58913876
Maybe it should be called RZ7-1700x or something, so that it doesn't get mistaken with the graphics cards. It looks kinda ugly though.

>>58918106
R4, R6, R8 would have been nice and simple while also making it obvious they are not graphics cards by being even numbered instead of odd.

>>58916598
GloFo's 14nm yields at the die sizes AMD is producing is in fact phenomenal. I doubt there will be shortages of the chips at launch, and if there are it'll be entirely because people are consuming them at an absurd rate.

>>58912654
>>58912712
I picked up a W541 thinkpad with the i7-4810MQ w/ ultra dock and 170w PSU brick 16GB ram for $500 cash. Looks mint has Sammy EVO SSD's and a DVD writer I can swap in and does m.2. Nvidia Quadro and has two years left on warranty under my lenovo account. Running legit Windows 7 pro not that Win10/8 trash. Actual quad core and kind of a desktop replacement (close to it). Runs gud. I'd jump on the Haswell's while they are still around if you can. It beats those shitty dual core i7/i5 laptops.