Specifications of Ryzen launch SKUs likely leaked

File: soon.jpg (76KB, 550x367px) Image search: [Google]

76KB, 550x367px

>>58758645
Translated:

Octocore #1:
base 3,65 GHz, max boost for 1 core: 4,0 GHz. 8 cores, 16 threads

Octocore #2:
base 3,40 GHz, max boost for 1 core: 3,8 GHz. 8 cores, 16 threads

Sexcore #1:
base 3,30 GHz, max boost for 1 core: 3,7 GHz. 6 cores, 12 threads

TDP likely 95 W for all.

>>58758645
Neat.

>>58758702
What sucks is no 4C/8T being sold early on, I guess.

>videocardz.com

>>58758740
They are actually very good for GPU news. The guy is sane and can tell bullshit fakes from scoops, unlike other graphics card news sources. If you check the track record, you'll agree.

His contacts probably grew enough to catch miscellaneous stuff in his nets, he leaked Ryzen/Zen slides ahead of earlier reveals too.

>>58758645
F4 OPNs are the latest stepping, latest microcode, but that doesn't mean their clocks necessarily reflect consumer products.

The 6 core parts are likely 65w.
8 cores hitting 3.6ghz at 10w per core plus 15w for uncore = 95w.
6 cores at 3.3ghz, even if they were still pulling 10w each, plus 15w uncore = 75w.
Two L3 slices are disabled, corresponding to each core, so power is saved there, and its extremely unlikely that the discrepancy in clocking is that large. 65w-70w is where this would land. Giving it a 95w TDP, which is thermal margin, then limiting clocks to a point where it would be impossible to utilize that thermal margin is something that wouldn't be done.

>>58758833
>that doesn't mean their clocks necessarily reflect consumer products
The clocks are quite high so I expect they will go with them for final launch.

The hexacore doesn't really need to have fitting TDP, they can just overstate it so that they can later sell some other 65W model at premium.

In any case, there will likely only be 35W, 65W and 95W SKUs, so with power at 75 W, it has to have 95W TDP.

>>58758862
It wouldn't be 75w. 75w is just the estimation assuming uncore is drawing like it would on a full 8 core chip, which isn't happening, because each core is tied to its L3. When a core is disabled a slice of L3 is, and power consumption drops along with it.
Giving a chip a 95w TDP when its only pulling 65-70w does absolutely nothing. The chip isn't pulling anywhere near its power limit. Its running cooler than all other comparable 95w chips so its not anywhere near its thermal limit.
The only way it could be justified as a 95w SKU is if it needed even more voltage for even lower clocks than the 8 core parts.

>>58758944
Chip doesn't need to use its full TDP. TDP is just liit or cooler/VRM rating.

The chip won't have higher power consumption just because you raise TDP. Well actually it could if that causes it to turbo boost more. Or if they give it shit voltage so that they can harvest more bad chips. AMD does that some times.

>>58758968
TDP isn't an arbitrary figure. All AMD chips know their set TDP because of how they sense and measure TCTL/Thermal Margin.
They wouldn't have a full 95w TDP.

i hope AMD has some sort of similar front end for manipulating ryzen overclocks and power states similar to wattman

>Engineering Sample
Who says these aren't old engineering samples or Naples?
Launch is in 4-6 weeks and Lisa promised widespread availability from day one. They must have final product by now, not engineering samples.

File: C1aelJ4UUAEG0za.jpg (150KB, 1360x1020px) Image search: [Google]

150KB, 1360x1020px

>>58760013
Everything not marked as a for sale part is an engineering sample.
The OPN end with F4, those are the newest steppings. The Ryzen test systems at CES were 3.6ghz base clocke and 3.9ghz turbo, and those were F3.
The top end F4 brought it up to 4ghz turbo, still 3.6ghz base.

Isn't the Turbo supposed to be variable based on temps and power limits? Is 4GHz the maximum Turbo or is it the "average boost clock" like they advertise on GPUs?

>>58758645
>higher core count = higher clock speed
????

Wow, that's pretty sad, only 3.7 turbo? Intels have no trouble getting to 4.5 nowadays..

AMD has given up, no point waiting for Zen for the smart consumer anylonger..

>>58762274
even my haslel i7 runs 4.6 @1.270v 24/7 for almost 3 years now

File: 1483193560378.jpg (66KB, 568x612px) Image search: [Google]

66KB, 568x612px

>>58758645
no 4c/8ht = in trash it goes, it won't win vs even i5 in single core perf, also bad overclocking potential with as many cores. Why amd must ruin everything i'm now on trash old 4core athlon x4 and i can't play anything since my i5 2500k broke. Now i will have to wait year or two for new Intel cpus since 7th gen is trash.

>>58762303
Neat, how much did that X99 motherboard cost?

File: 1483628859502.jpg (156KB, 633x758px) Image search: [Google]

156KB, 633x758px

>>58762274
it's fucking over amd is finished, even i5 6600k has 3.9mhz turbo... and i'm not even talking about i7 or 7th gen of intel cpus which all are over 4ghz.

>>58758740
They are pretty credible compared to other shitty rumour sites. Most of their info on rx480 were true.

>>58762274
This, the company is done for I have NO idea what htey're doing anymore

well they say the product reflects the company shit product = shit company

>>58758645
Ryzen 3.8 Boost Clock +OC might be gud

@58762274
@58762303
@58762314
@58762329
@58762336
not getting (((you)))s dumb fucking kids

>>58758645
8c/16t at that clock is fucking good, if they don't fuck it up asking 1000 € I am in.

>>58762389
Nobody cares about 8 cores, Intel has a 10 core for people who actually need cores.

>>58762399
a 10 core at 3ghz for more money than you earn in a year, shlomo

@58762274 (not you)
>Turbo frequency means maximum clock speed

But that's where youre wrong. Even my old 4130 could clock 5.0 with an aftermarket cooler. Its turbo was something like 3.8.

AMD has a record of having all of their chips unlocked (unlike Intel) and people have gotten some pretty batshit clock speeds with crazy cooling setups (Stable 8.4GHz with liquid helium as your coolant, anyone?)

>>58762399
8 cores at 3.6GHz(unknown turbo, possibly 3.7 or 3.8 from CannardPC test) would obliterate a 10 core 3.0GHz(Intel turbo doesn't kick in when a lot of cores are active) unless software core scaling is 80%, which is damn near impossible without some _VERY_ specific applications

>>58762451
99% of AMD cpu buyers are budget gamers so it would be worthless cpu for them if your single core perf will be trash and you would be oc that. They would be better with 4c/8ht if it means higher oc potential and perf in games would be the same.

>>58762494
and you couldn't oc that as well as intels i5/i7*

>>58762494
Quite an assumption.
Maybe it's projection? Or just a small hope?

>>58762494
I'll just remind you that the most powerful TOP500 machine in USA is running 18,688 Opterons.
https://en.wikipedia.org/wiki/TOP500#Top_10_ranking

So no, keep your projection to yourself you dumb underage /v/irgin

>>58758740

surprisingly enough, it's one of the few websites that actually discuss the things they claim to do

>>58758645

Any info about price points? Will Ryzen be reasonably priced? Or will they go full Intel with sub-par performance?

File: Another Major Disappointment.jpg (12KB, 495x91px) Image search: [Google]

12KB, 495x91px

>>58762506
>>58762537
i don't give a fuck about non-gaming cpu use, majority of people WILL ONLY BUY THEM TO PLAY FUCKING GAMES YOU MORONS if i'd like to do calculations i'd use calculator

>>58758678
>sexcore

>>58762537

Completely irrelevant

@58762568
I think a know a place that's more your speed kiddo!

>>>/v/

>>58762568
Plenty of people care though, just because you're underage /v/ermin doesn't make your opinion worth a thing.
It's better to sell 1000 $700 CPUs at 90%+ margins than 3000 $150 CPUs at 40%+ margins I'll let you figure out why that is.

File: kill_yourself.png (86KB, 1571x230px) Image search: [Google]

86KB, 1571x230px

>>58762274
>>58762314
>>58762329
>>58762336
>>58762399
>>58762568

This isn't even funny anymore. Even the shilling on /g/ is shit now.

>>58762597
Right only poor GAMERS buy AMD them GAMERS amirite

>>58758678

What happened to the 4c/8th version (~50W). Is it still going to be released?

>>58762657
If there is gonna be it's gonna be in low supply due to how binning works.

Hey all, I don't know where else to go so here goes.
My silly brother installed Lucky Patcher on his device and now he is afraid after seeing its folder name which reads com.android.vending.billing.InAppBillingService.LOCK
Is this a normal directory for Lucky Patcher stuff? I haven't used it before so there's no way I can help him. Can he just uninstall the app and the folder and its content deleted as well? He said he launched it but didn't use it. Thank you for your time and explanation.

>>58762657
Not as a part of the Summit Ridge lineup, the Raven Ridge coming out later this year is their pleb tier 4c/8t.

>>58762504
Reminder that if a core iX series isnt a K model, it's not unlocked and you can't overclock it. Like >>58762449 said, AMD usually unlocks all of their CPUs

>>58760050
I think it is actually 3,65 GHz base. Ryzen can clcock in increments of 25 MHz, but their naming system for ES chips didn't count with that.

The detected clock of 3640 MHz would match that, look how the other chips are reported as 3290 and 3390 MHz.

>>58762657
They will probably also make quadcores from defective octacores, but they'll release those at a later date than Q1 2017.

Bulldozer and Vishera also had 4core variants - I don't recall if it was released immediately or with some delay, though.

The quadcore will likely be rated at 65W, that is what the earlier ES chips had as TDP, even if real power consumption might be lower.

>>58758678
Where are you getting that information from?

Intel alternative news factory?

>>58763038
Thu turbo boost clocks are encoded in the ES names, as well as the base clocks (see the 37/33, 38/34, 40/36 parts). This has been known and confirmed for many years and Ryzen ES naming matches this system.

Also see the detected clocks in videocardz's information - those probably come from some OEMs like Maingear/Falcon Northwest who will release gaymer rigs for rich kids with these CPUs.

>>58763070
>max boost for 1 core
Where's this information from?

>>58763114
Well, that is not known for sure, but based on current working of AMD CPUs and the information about the first ES chips that leaked in july or august, the maxmimum turbo boost is not available when all cores are active.

There is reportedly another clock set as maximum boost for situations where all cores are loaded, it would be somewhere between the base and the maximum.

But there is that OC boost thing too, where thes chips can allegedly automatically turbo beyond the limits if they have good cooling, so the limit might not be hard and unbreakable.

>>58763114
>>58763178
See this post: https://forums.anandtech.com/threads/new-zen-microarchitecture-details.2465645/page-89#post-38363321

This guy's information was proven correct few months later, so he was most likely a legit leaker.

The clocks have improved a lot since, but there is probably still a separate max "all core turbo" and "one core turbo". Well, it might actually be a 2-core turbo or half-cores-turbo, not necessarily single core only.

>>58763205
>>58763178
So rumors and speculation, not actual facts.

>>58763276
His ES information was later confirmed, the codes appeared in the AotS database.

It's not 100% proven facts with zero possibility of being wrong (those will only come when it is released), but not rumours or speculation either.

>>58763293
Let me correct myself.

Rumors and speculations based on early engineering samples.

Not facts based on retail SKUs.

>>58763396
Yes, it is possible that it can change, but that is less likely course than it staying the same, IMHO. So for the moment I only count on the more restricted turbo mode.

But there is that "extended frequency range" thing too.

Just one month. Will you all calm down a bit?
AM4 boards look good, asus even used "value" term on them meaning it's no where near z99 prices.
Even if it is $400 it's still will be cheaper than 6800k.
Early March wide availability.

Well it's a common platform top to bottom. You'll have silly plastic&LEDs and sthit stuff for 300 USD, and also cheapo mobos for 60-100 USD.

For huge overcocks you might need ones with above-average VRM design tho, because the lower voltages on 14nm mean higher currents to satisfy the same CPU TDP.

>>58763528
cheapest x99 is $230

>>58763595
Yeah, which sucks if you want just a simple PC, don't plan on OCing, but want a better CPU than i7-7700K for work purposes.

AM4 will be nice in that.

>>58762578
sexcore will be cheaper than octa

>>58762274
>why must an older, lower-tier CPU be slower than a newer, higher-tier one

I wonder if they'll make a tricore in the APU range. I always liked that. Crippled chips like that are moe.

>>58765499
Some defective quadcores might get sold as 3c6t.

It would be pretty interesting

>>58765511
3C/6T could be enough for my purposes, all I want from it is being able to decode 4K AV1 or 12bit HEVC or whatever format will come next.

>>58765499
Rumor says they can disable individual cores, so a 3 core is possible. The 6 core is two modules of 3.

>>58765768
Yeah, based on that on tricore should be possible on APU (has just one CCX).

File: lvl1okhand.jpg (16KB, 209x230px) Image search: [Google]

16KB, 209x230px

>>58765327
it's spelled SIX in english, or hexa if you wanna go with the same theme you used for the 8 core (octa)

>>58765916
I'll sacrifice octa to have sex, thanks you.

>>58758645
>no 4C8T
>6C8T locker clock than entry 8C16T
Fuck you AMD.
If this is true guess i'll get a G4560 until i'm trough college and call it a day.. fuck this.

>>58766397
You don't want to give lower-tier chip higher clock, so that it doesn't end up faster than the higher-core number chip.

That is the reason why i5 always has lower turbo, to give i7 some exclusivity advantage.

You are gonna OC anyway, aren't you? Lower starting clock you will get it for less money and will get higher relative overclock for free.

Hmm, I just saw a retail/middle-level distributor guy subtly suggest that the hexacore chip might be unavailable for some time, or really hard to get (like RX 480 4GB?) right after the launch.

>>58766599
AMD doesn't make 6 core silicon, they need defective 8 core silicon to re purpose it for a 6 core one.

defects are a part of yields, and if yields are very high there's not gonna be enough defective 8 cores to make 6 cores, unless you're purposely lasering off cores on the 8 core to make a 6 core, but that's unprofitable and just a waste of good 8 core silicon.
This is also why I believe there's not gonna be any 4 cores from Summit Ridge.

>>58761898
I think you hit the nail on the head

File: 16473561_1899772306914352_4038126753597289275_n.jpg (109KB, 960x926px) Image search: [Google]

109KB, 960x926px

>>58762676

Wtf? Wrong thread, anon.

File: AMD%20Zen%20December%202016%20Update_Final%20For%20Distribution-page-017.jpg (428KB, 3999x2250px) Image search: [Google]

428KB, 3999x2250px

>>58761898
Advertised Turbo is never maximum Turbo, neither for Intel, nor AMD, or Nvidia.

Intel Turbo 3.0 with Broadwell-E can hit 3.9GHz on a single core, it's rated at 3.7GHz.
Nvidia GPUs can turbo over their rated turbo frequency depending on power and GPU utilization.
AMD is the same with Zen, they explicitly stated that the Turbo can go over the advertised limit with XFR.


What these companies never say how big is the Turbo for multiple cores, because there's no way in solid hell you'll see the 8 core Intel turbo all the cores at 3.7 or AMD turbo all 8 cores at 4.0.
Unless their TDP limit is 200W, of course.

>>58767074
Or unless the you enable it in your bios like a regular person

>>58767086
The regular person doesn't even know what the BIOS is.

>>58762675
It will be in high demand because of the price

>>58767074
Current AMDs require half of the cores being put into a sleep C state, before the rest can hit the highest turbo bin.

>>58767097
That turbo shit shouldn't be on with an OC anyway

>>58767107
You don't understand, due to how foundries work it would cost AMD MORE money to make quad core than a 6 core.
4 cores won't be in supply because they need to stock them up, from a single wafer you can get around 260 chips depending on approximate die size, and lets say the yields are around 50%, so 130 chips.
Out of 130 of those chips, the large majority will be 8 cores, the minority will be 6 cores, and the leftovers of that 6 core minority will be 4 cores.
AMD needs to stock up on these 4 cores before it can put them on sale.

>>58767208
Or it can just launch the 4 cores without any supply, piss off everyone like with Polaris/Hawaii and Llano in case of Apple.

>>58767295
Or skylake

>>58767208
Maybe quadcores will just be APUs with broken graphix :^)

>>58767356
The APUs "raven ridge" is set to launch in summer, so you're still off waiting a while.

>>58767295
There's no point of artificially disabling half of the die. They have old AM3 and FM2 stock to clear until they will launch 4 core APUs with Raven Ridge in Q4

>>58767378
Exactly, I can only see high availability 4 cores if people don't buy their 6 cores for whatever reason.
That or if yields are fucking terrible, as in Fermi terrible.
There's just no natural way to get a lot of quad cores from Summit ridge.

File: 1480764634428.png (11KB, 306x310px) Image search: [Google]

11KB, 306x310px

>>58767378
>They have old AM3 and FM2 stock to clear

good luck with that

>>58767378
The will most likely sell a Summit Ridge 4 core, but not immediately.
Partly because they don't want reviewers to focus on the 4 core.

https://www.techpowerup.com/230291/amd-readies-ryzen-platform-drivers-for-windows-7

>>58767779
Fucking awesome

Kaby lake btfo

>>58767779
>That dinosaur of operating systems should be already discontinued. Who wants to run hot new hardware on a 7½yr old OS? Have fun playing DX12 games, oh wait.. :rolleyes:
lol

>implying you need dx12 with vulkan

>>58767779
ITS OVER AMD IS FINSIHED

>>58767779
No surprise. AM4 is already shipping with Bristol Ridge and Windows 7. No reason for AMD not to support Windows 7.
No idea why Intel doesn't support Windows 7.

>>58768019
>No idea why Intel doesn't support Windows 7.
Microsoft asked them not to so people adopt W10 faster.

>>58768019
microsoft are jews

>>58767630
Unlike intel they're going to discount shit

>>58758678
>TDP likely 95 W for all.
This is quite interesting for me if it's actually true (and accurate in terms of real-world values). So we're potentially getting a 6C model with lower clock at the same TDP as the 8C versions, this would mean that the 8C CPUs are getting much better silicon. I wonder if we will end up with 8C CPUs OCing very well.

>>58768758
Or the 95W for six cores might just mean AMD is giving you ample room to overclock.
But there's no doubt the 8 cores will be getting better silicon, but not much better.
That's for enterprise customers.

>>58767893
>implying you need dx12 with vulkan
most commercial devs hopped on the dx12 bandwagon already. there are like 5 studios using vulkan.

this ship has sailed. nothing changes. cancer wins.

>>58767779
>Now someone seems to have sand in a wrong place, or something similar..
>I don't know about you, but the main priority about an OS isn't the looks for me, I could use that W2K theme without problems.
>Spying me? I don't give a crap about that, why I should care what they do with a random guy's information? I don't care if someone thinks "lol, this dude seems to like cat videos" or something like that.
>Also, I don't care if the OS installs updates and reboots when I'm not using my PC.
>Always on Windows Defender isn't also a terrible thing, since why I should install a third party antivirus?
>I really recommend you taking few deep breaths, if a TPU comment makes you rage that much.
>E: Trolling? Haha, sure. I installed Win10 right after it was released, and there is absolutely no reason to use any older version of Windows.

Really fires up the neurons.

>>58768777
That's why I said it's interesting if the number is accurate. If it's actually a 70-75W CPU with a 95W label on top it doesn't mean anything and silicon quality between 6C and 8C parts could very well be similar.

File: 1471561119231.jpg (71KB, 552x661px) Image search: [Google]

71KB, 552x661px

>>58758645
Fuck it's DoA. Can't believe I was actually waiting for this.

>>58768777
>That's for enterprise customers.
That's why I'm waiting for workstation/enterprise-centric Ryzens for the best binned CPUs and motherboard. I think none of the AM4 chipsets will work with ECC RAM out of the box.
Hopefully Ryzen will have sub-45W TDP eight-core variants so I can move on from my Xeon D-1521.

>>58768871
Hey, Xeon D is Intel's most interesting CPU in the last 5 years.
I don't know why you'd get rid of it, a lot of cores, with plenty of I/O.

>>58768898
>I don't know why you'd get rid of it
Four cores at 2.7GHz maximum can only do so much

>>58768927
Obviously it's not a gaming CPU, but you can put your entire NAS/DHCP/mail/DNS/VPN/router server on it and it won't bat an eye.
Those 2x 10GbE links aren't just for kicks, it's a CPU with a specific purpose.

>>58768898
>>58768871
There was a rumour floating around that zen two die core modules will all be embedded parts to compete with xeonD parts. Whilst all socketed parts are the four die monsters.

>>58768956
>NAS/DHCP/mail/DNS/VPN/router server on it and it won't bat an eye.
Wrong, the four core version I have isn't strong enough to deal with DNs+VPN+Proxy+IPS. It's the IPS function alone that causes it to shit the bed and throttle LAN speeds to just under 400MB/s.

>>58768981
Why is it choking with IPS? That should be handled with GPUs if you have a lot of cameras around.

>>58769003
>That should be handled with GPUs if you have a lot of cameras around.
Niggawatt
I use Squid, which is still mostly single-threaded and loves high core frequency. I didn't even enable all of the detection rules.

>>58769046
We were talking about a different IPS.

https://en.wikipedia.org/wiki/Intrusion_prevention_system

>>58769068
And that's the IPS I'm talking about. I use Squid because it's the one I'm more familiar with.

the wait is killing me

>>58769102
So a 2.5GHz XeonD is insufficient?

Squid should have been reworked to use some SIMD like AVX if it loves high clocks that much.

So will the price of the FX 8350/8370 dropped immediately after Ryzen is available, or will I have wait a few weeks?

Only asking just in case Ryzen is shit and DOA or Good but not worth the price.

>>58768795
>most commercial devs hopped on the dx12 bandwagon already.

But source 2, unreal engine 4, and unity have vulkan support

Android ""games"" are huge as fuck these days

File: 1483237996531.jpg (275KB, 2480x3507px) Image search: [Google]

275KB, 2480x3507px

>>58758645
This is absurd
No one cares about your GHz obsession

We want prices

>>58769660
Prices will probably be kept secret or "to be determined" till the last second.

>>58769159
It's already quite low so idunno if it will go down much. There will probably hardly be an overlap, but maybe some 10-20 % discount could happen. They however tend to appear 4-6 weeks before launch of new generation.

>>58767779
HELL YES, FUCK YOU INTEL.