memes aside, is ryzen actually good? there are a lot of benchmarks

>>59691903
>also would it be dumb of me if i were to replace my 4790k with a 1700x?

Depends on your workload, buy hardware for your software.

>>59691926
i game and stream (gpu encoder). im mostly interested in changing the cpu because i want to, not really because i need to.

File: Screenshot_20170401-155916(1).jpg (210KB, 720x733px) Image search: [Google]

210KB, 720x733px

It really depends what you intend to use a cpu for.

Personally getting R7s is pretty stupid if all you want to do is play vidya gaems and browse the web. R5s are better for that but will come out later.

Anyway as it stands right now unless you're on a haswell or later motherboard there is no point in buying intel housefires anymore especially delidalke.

>>59691903
it's great for meme'ing and has the great benefit of making intel shills rage

>>59691903
Yes, it would be dumb to replace your cpu for less than 50% performance gain.

>>59691903
It's good as long as you buy the right RAM. Fast RAM is crucial to its performance, but it can only boot newer single rank Samsung B-die stuff at 3200MHz+ right now. The cheap dual rank Hynix trash used in most DDR4 to date is stuck at 2933MHz at best (more likely 2666MHz for most boards). The difference in RAM speed is why benchmarks are all over the place.

For what it's worth, I'm liking my 1700, which is the only one worth buying since it overclocks just as well as the more expensive ones (silicon lottery permitted).

>>59691960
>Game and stream
Upgrade. Despite all the memes, multi core AMD handles this better than Intel.

>>59691990
What voltage and frequency is yours running at?

>>59692014
4GHz, 1.35V set and 1.356V actual with all cores loaded. ASRock Taichi board with 2666MHz RAM (since I have cheap Hynix trash from a previous buiid). Stable in all my testing so far. I know others can only hit 3.9GHz at similar voltage though.

File: 03_big_0.png (49KB, 849x621px) Image search: [Google]

49KB, 849x621px

>>59692052
holy shit, I hope I can win the silicon lottery like you did. I'm planning on getting an R7 1700 too.

>>59691972

Fucking schills can't let AMD win anything can they? Deliberately underclocking the intel chip so it can be at the top of the graph. They know exactly what the fuck they are doing too.

>>59692152
Yeah it does feel like that's the case unfortunately, ryzen has had a ton of success undervolting but it wasn't in their best interest to show that.

Oh well at least that graph can put the AMD = housefire meme to rest. AMD has come a long way from the faildozer days.

>>59691972
After checking benchmarks the 1700 is barely faster than my stock i7 2600k in games

File: tvtkbtb.jpg (395KB, 2400x2000px) Image search: [Google]

395KB, 2400x2000px

>>59692268
>After checking benchmarks the 1700 is barely faster than my stock i7 2600k in games
It's getting better as winblows/bios/microcode updates get passed on. Unfortunately ryzen was not as good of a launch as everyone expected on day 1.

Anyway if the only thing you do is vidya then upgrading from your i7-2600K doesn't seem like it should be done at all. A GPU upgrade would be better.

>>59691903
I think Vega will tell the real answer to this. People have now been finding that NVidia drivers are causing issues with threading and thus limiting R7 performance.

I think upgrading to R7 might be a bad idea for you, but it is better than upgrading to a 7700k, if only for the fact that you can keep your AM4 mainboard when zen2 comes out. There are also still a lot of updates coming that will help the R7's perform better.

>>59692313
except the min fps though.. ryzen supposedly delivers much smoother framerates and frametimes overall.. i'm chaing xeon 1231v3 to ryzen 1600X when it arrives in 10 days

Ashes of Singularity performance boost from the Ryzen optimizations was pretty amazing.

op here, whats the best mobo to go with an 1700x? and whats the best ram selection for said mobo, because i know gigabyte currently has an issue with that, so im expecting others do, too.

>>59692644
If you are not planning to SLI, choose a B350 board, it will be much cheaper, and you can still overclock

hey. im running a xeon e3-1231v3 and a gtx 1070. i thinking about to upgrade my xeon. i do stream sometimes and playing some video games

>>59692644

They're all pretty much on par now. I would pick a good B350 or a budget X370 chipset.

http://support.amd.com/en-us/kb-articles/Pages/SystemBuildingFAQ.aspx

Read that.

There's also supposed to be a RAM update in April that will allow broader RAM selection, as well as auto overclocking XMP profiles.

Happy building.

>>59692644
Don't buy a 1700X unless you're not planning on overclocking. The 1800X is the only one which has gains in maximum frequency on average, and it's to the tune of 100-200MHz (i.e. not worth the outlay). The 1700 is by far the best deal.

Best board is either the ASRock Taichi or the Crosshair VI Hero, though you don't really NEED to pay that much. A B350 board will be fine for most people. Just avoid the ones with no CPU VRM heatsinks.

>>59692688
For streaming it'd be a large upgrade. For gay men, still a decent one so long as you're overclocking past the Xeon's frequency.

>>59692713

Wrong link, my bad.

https://community.amd.com/community/gaming/blog/2017/03/14/tips-for-building-a-better-amd-ryzen-system

Is it worth the upgrade from an fx8320? Or does my CPU still have some years left? Games and video editing

>>59692963
It's a huge upgrade for both. You'd see far higher framerates in gaming (especially minimums) with any decent graphics card.

>>59692983
Just recently upgraded my 270x to a 1070. Maybe about a month ago so right now I'm stuck on weather to upgrade Mobo, ram, and cpu

>>59692963
wait for zen+ or 2 or w/e it'll be called if you don't already hate your life
but it's a good upgrade both ways

Deus vult

>>59691903
if you only game and do literally nothing else on your computer, intel. Almost anything else CPU intensive ryzen kicks ass. Code compilation, encoding, rendering, compressing, encrypting, etc. benefit tremendously from the extra cores. Also if you expect overclocking go intel K series, ryzen will not go past 4Ghz without enough volts to insta-fry your build

>>59693010
Then you should absolutely upgrade. An 8320 will massively hold back a 1070 in most games. Even in the heavily-threaded ones where Piledriver can shit, Ryzen will shine even brighter since they're perfect for it too (and it has far better IPC).

>>59693058
>where Piledriver can shit

*shine

Freudian slip.

>>59691903
>is ryzen actually good?
https://www.youtube.com/watch?v=TcdmeGOsnss

You tell me.

>>59692100
The colour coded numbers are the CPU temperature right (in *C)?

(volts on the left, MHz on the top)

>>59693147
No m8, it's watts. I don't think any desktop cpu can even operate beyond 100C.

File: 1490693202396.jpg (40KB, 470x469px) Image search: [Google]

40KB, 470x469px

>>59692268

>i buy a 8 Core CPU with 16 Threads to play fucking video games

>>59693125
The real reason for poor benchmarks in games is Nvidia's shitty DX12 support is holding it back.

http://www.youtube.com/watch?v=QBf2lvfKkxA

http://www.pcworld.com/article/3175775/computers/ryzen-7-1800x-and-radeon-fury-x-building-the-water-cooled-fire-breathing-apex-of-amd-power.html?page=3

http://www.youtube.com/watch?v=0tfTZjugDeg

>>59693202
I've seen CPUs get posted at ~98*C so that's why I was wondering.

But good to know.

Thinking to update my Mobo-CPU too the AM4, as this old Intel of mine is shitting the bed constantly on a core error as of late.

>>59692713
Asrock AB350 Pro4 or MSI B350 Tomahawk? Does the extra price justify the Asrock one? I've heard the MSI B350 Gaming Pro is about on par with the Tomahawk too except for having fewer expansions slots but still can support 3200MHz memory, and it's like $60 cheaper (in my third world currency).

>>59691960
Don't use a GPU encoder for streaming, they need way higher bitrates to perform well.

>>59693316
>it was nvidia's drivers' fault all along
You couldn't make this up.

>>59693461
The MSI will be fine. You won't get 3200MHz RAM running on it unless you buy a CL14 Samsung B-die kit though.

File: OjSIcgM[1].jpg (387KB, 2201x1071px) Image search: [Google]

387KB, 2201x1071px

>>59693517
More proof?

File: Hh02eaV[1].png (15KB, 1027x208px) Image search: [Google]

15KB, 1027x208px

>>59693588
Furxy X on Ryzen vs Titan X

File: 1487730850625.jpg (20KB, 480x360px) Image search: [Google]

20KB, 480x360px

>>59692313
>mfw this is identical to 1700 at 4.0GHz (in terms of silicon lottery it's about as good as 1700X)
>mfw I can get a future proof CPU that runs better than 7700K for $320
>mfw the 1600X will run the same for $250

>>59693588
Everyone knows your dated nazi benches dont mean shit after optimisation including recent windows update

>>59693588
>>59693608
The are pretty hilarious. Never thought I would say the day the fury x was on par with a titan x.

>>59691960
>i game and stream (gpu encoder).

>>59693485
>Don't use a GPU encoder for streaming, they need way higher bitrates to perform well.

/thread. ReLive and ShadowPlay have really low overhead, but their quality/bitrate both rather suck. Having more cores to throw at h.264 software encoding is half of AMD's argument for Ryzen gaming.

>TFW you bought a ryzen 7 1700 at launch but you still can't get your hands on an x370 taichi.

Provided you don't plan to overclock much, exclusively play games without a secondary workload, and are OK with the Early Adopter tag, yeah it's pretty alright.

Anyone can tell you these are bad for gaming, even though they say that because the red bar wasn't as long as the blue bar and don't consider that the numbers are still really impressive.
And before anyone calls me an AMD shill - I have an 5820K. Not out of want, mind you, but necessity. AMD hadn't done a damn thing worth my attention up until now. Ryzen's a legitimately good thing, and whilst it's not a record breaker in performance...the fact that you can get $1000 performance for $500 now (or even less if you're willing to OC) is what's paramount.

If I buy 3200 Mhz ram from the qvl list of a motherboard supporting 3200 Mhz, an I 100% guaranteed to hit that speed today?
And if I OC the CPU that guarantee is gone?

>>59691960
Don't, Ryzen is meant for multithread load. Or games that need more than four cores, once that time comes. It might work as ahead-of-time upgrade if you like shopping, but just for games, I don't think you need to upgrade yet.

>>59691903
R7 1700 OCs to 3.9GHz average and 3.8GHz on stock cooler.from 3ghz base.Performance is on par or better than the 6900K when OC'd and paired with decent ram.
The lower end Ryzens completely depreciate the i5 and i3.

>>59691903
heh. i'm no gaymen faggot, but:

-the 1700 goes for 430 dollars the cheapest
-cheapest am4 mb goes for 160
-plus i would have to get a new graphics card, and the crappy ones like the 1050 goes for 100 over retail at 270, the less crappy ones like the 1070 is above 600

on the other side

-the 7700k goes for the same 430. the 7700 is around 380
-cheapest mb is around 90
-no need for a graphics card if i use the integrated one.

easy decision. gonna stay on my pentium 4 until it dies.

>>59693255
>tfw I fell for the i7 meme
I was young and foolish

File: photoshop.png (28KB, 500x450px) Image search: [Google]

28KB, 500x450px

>>59691903
Ryzen is truly horrible. It's literally only useful if you do rendering/encoding or similar tasks all day long. It sucks at basic tasks like productivity and gaming.

>>59694232
|
|>
|
|x
|
|

File: ryzen-streaming.png (129KB, 1440x1080px) Image search: [Google]

129KB, 1440x1080px

>>59691960
Ryzen sucks at streaming too

File: ryzen-cpu-value.png (26KB, 773x371px) Image search: [Google]

26KB, 773x371px

>>59694243
AMD is jewing you hard with Ryzen, you literally pay twice as much for worse performance

>>59694251
> tfw Ars probably used software encoding settings that look worse than ShadowPlay

>>59694243
Brainlets like you who fall for such shitty bait are the real cancer.

>>59694265
Yeah I mean if you used GPU encoding it would be even more of a slaughter because the Intel CPU is way faster in gaming

>>59691903
There's no reason to upgrade a 4790k to anything, whether it be amd or intel. Atleast if your gaming/streaming. That said the ryzen cpus are good, now that they've gotten better ram support and bios updates, theyve seen significant improvements. However they're still plagued by those outdated week 1 benchmarks that people continue to throw around.

>>59694274
>benchmarks are bait

AMDfags are so delusional

>>59693838

Yes to the first. But you have to manually enter RAM timings and voltages.

No to the last. Overclocking your CPU helps overclock your RAM. Because the memory controller is integrated on the CPU die.

>>59694322
>But you have to manually enter RAM timings and voltages

Most QVL kits should boot via XMP alone. I've seen plenty of people report success with just doing that with the right kits (CL14 3200MHz stuff or anything 3600MHz+). It's generally when you buy something not on there (and not single rank Samsung) that manual tuning is necessary.

File: 1700X 7700K 1080p crysis 3.jpg (423KB, 1920x1080px) Image search: [Google]

423KB, 1920x1080px

https://www.youtube.com/watch?v=0tfTZjugDeg
>proof that shitty nvidia DX12 drivers are gimping ryzen on RotTR
>mfw throttlefags are still trying to defend their aged quad core vomit

>>59694322
>>59694374
Neat, but how can you tell if it's single rank with the heatspreaders on there?

>>59694533
Any CL14 3200MHz kit and any 3600MHz+ kit will be. They're all Samsung B-die chips, whereas kits with looser timings or slower speeds are dual rank Hynix stuff (which Ryzen doesn't like at all right now). Anything which fits either of those conditions is safe.

>>59693818
Did you pre-order your taichi board? I got mine from Newegg, but it took about two weeks to fulfill the pre-order. So far running pretty well on the 1.60 bios with a G.Skill RipJaws V 16 x 2 @ 2400 kit on the built-in XMP 2.0 profile with no problems. The board did boot up the first time with the p1.40 bios with that ram kit at 2133 stock letting me update first to the 1.50 from a dos image, then through the bios flasher to 1.60.

>>59694322
Would a G.Skill Trident Z 3200C16D be compatible enough with a good B350 mobo? The one with CL14 seems more likely to reach 3200MHz but I don't feel like paying 40 bucks more for it.

>>59694576
Thanks.

>>59694682
Go to the motherboard website or the ram vendor and check the qvl list for compatibility.

>>59694576
Does this only apply to sticks that are 8GB or less?

>>59694682
You won't get anything more than 2933MHz out of that right now, and more likely 2666MHz. AMD say they're working on RAM compatibility though, so it might pay off in a couple of months.

>>59694736
The new Flare X series from G.Skill has 16GB single rank sticks at 3200MHz.

https://www.newegg.com/Product/Product.aspx?Item=N82E16820232530

>>59694889
>The new Flare X series from G.Skill has 16GB single rank sticks at 3200MHz
The link you gave is to a 2x8GB kit. Are they available elsewhere, or just not quite yet?

>>59694913
Oh yeah, the hazards of posting at 1am. In that case no, just 8GB sticks for now.

File: 1490920686398.jpg (112KB, 1024x934px) Image search: [Google]

112KB, 1024x934px

I have a 4690k at 4.6ghz, I mainly play games, but I despise hsving to exit out of all my programs so it isnt a stutterfest, would ai benefit from Ryzem?

>>59695128
A 4690k will be good enough for at least 3 years for games only, but a 6+ core Ryzen should help a fucking lot with the stuttering issues. For 220 bucks you can get a R5 1600 which is easily the sweetspot for Ryzen.

>>59692268
Enjoy your outdated motherboard.

>>59692644
Get the Askrock Taichi and that G.Skill ram designed to Ryzen. I think it's 3000mhz 14-14-14 or something.

>>59691926
What about 3d modeling like rhino?

>>59693742
Only reason is the gpus aren't being stressed at 720p.

>>59693083
Thanks man! I guess I should ask for it at work, I think I'll buy part by part so it doesn't effect my wallet as much(single dad here)

>>59691903
Don't

>>59695731
I meant to say o.t

>>59691903
The reason why you are seeing conflicting benchmarks is because you are seeing them out of context.

If you want a real answer to your question then you will be better served to just go and read actual reviews by actual reviewers. Anandtech almost always puts out quality stuff. Arstechnica has a bias towards nvidia/intel, PCgamer is of course only looking at these cpus from a gaming perspective, etc.

The reality of ryzen lies somewhere in the middle. It falls short of the 7700k in 1080p gaming. However, it still is able to pull 100+ fps in most titles. More threads will allow you to do more while gaming/streaming/whatever with minimal performance loss.

>>59693899
>>59693058
>>59691972

These aren't really accurate. If you game at less than 144hz then a Ryzen CPU will give you comparable performance to a 7700k. Extra threads are always a good thing, even if you don't have a usage case where you "need" them, you will still benefit from having the extra multitasking capabilities.

4c8t is only the norm because Intel has kept the market in artificial stasis for 10 years. It is really difficult to keep recommending i7s because the market is clearly moving in a direction that favors more cores. It has been since athlon, core2quads, the i series, and now with ryzen.

To answer your question OP - if you are upgrading because you want something new, then Ryzen is the obvious choice. Don't expect crazy performance gains in games over your 4790k, you wouldn't see them with a 7700k and you won't see them with Ryzen.

File: fullretard.jpg (126KB, 450x373px) Image search: [Google]

126KB, 450x373px

>>59695627
>$200 ram

>>59695828
All RAM is fucking expensive these days.
Ryzen really benefits from fast RAM.

File: wd2-16.png (5KB, 515x230px) Image search: [Google]

5KB, 515x230px

>>59695793
>Ryzen CPU will give you comparable performance to a 7700k

That's hilariously untrue, look at the frame times, Ryzen is awful compared to the 7700k

>>59696106
>but muh unoptimized gamyen
Fuck off. Ryzen slaps the living shit out of everything Intel has to offer in raw performance. Buying a i7/Ryzen7 for gaming only is stupid as fuck anyways.

File: 85886.png (45KB, 650x350px) Image search: [Google]

45KB, 650x350px

>>59696991
It barely beats the 6700/7700k in encoding. And this isn't even the bad one. (it loses in h.265)
If AMD had priced Ryzen 7 at $200 it would be a clear winner. Priced above the i7 it's meh. I could've bought an i7 6700k 18 months ago if that's what I wanted.

>>59694251
>performance from before the 1st Ryzen update
Okay schlomo

>>59693255

>only playing games

What kind of scrub doesnt do other things while playing?

File: x265-2.png (582KB, 600x600px) Image search: [Google]

582KB, 600x600px

>>59697108
>H264
hello 2010

>>59697108
To be fair, aren't these benchmarks from a month ago though? There should be an improvement for the R7 with the release of Bios updates + faster memory speed.

>>59691903
Modern games are trash anyways.

File: 9 x264 Pass2_575px.png (142KB, 678x707px) Image search: [Google]

142KB, 678x707px

>>59697175
Encoding isn't really RAM speed limited.
(numbers are CAS)

>>59693255
lmao stuck in the past

File: UT3.jpg (25KB, 528x306px) Image search: [Google]

25KB, 528x306px

>>59691903
I'll close this down right now.
Ryzen is a mixed bag. It performs really well in the right situations. As good as kabylake, if not better at the same pricepoint. But there is software that either doesn't utilize it properly or runs into a kink somehwhere with the new Arch. Most of the recent large title games will probably see updates for ryzen, but the further back you go the more likely it is that there won't be any performance increases with the rare exception of pic related.

For those waiting for AMD to smash Intel in literally every use case, you'll have to wait for Zen+ and even then I wouldn't expect it. Look at AMD, their budget compared to intel, and then all the hurdles AMD has to overcome. The fact that they are where they are now with Zen is kind of impressive

That being said, you'll pay less and have a wider upgrade path. It APPEARS cannonlake will be on the same socket as kabylake but I'm skeptical since supposedly there is supposed to be a consumer 6 core offering. And no doubt that Icelake will be on a new socket. Meanwhile Zen+ and possibly Zen++ will be on the AM4 socket.

>>59697225
>x264
>>59697155

>>59693588
>>59693608
>Fury X 15% highest FPS than Titan XP
good1 nvidia!

good drivers these days as always

>>59694262
buy the 7350k!
better than 7700k!
7350k new best!

>>59695128
4c/4t have been starting to suffer for a while.

You could upgrade to the 4790k. That would significantly reduce your stuttering.

A 1500X should be better in most games, but will still be worse in a few, especially if you're using an Nvidia GPU until they fix their drivers.

File: 14899443954620.jpg (179KB, 1282x714px) Image search: [Google]

179KB, 1282x714px

>>59697357
>he said, full of shit

File: 287_S1Ep4.png (2MB, 1920x1080px) Image search: [Google]

2MB, 1920x1080px

>>59697381
> Nvidia GPU
>fixing drivers
You maki me laugh anon

>>59697391
see
>>59694262
idiot

>>59697403
you win

>>59693687
The 1600x is going to be the big seller imo.

>>59697381
1500x seems like the sweetspot for the 4c/8t line desu. I don't think anyone would get the lower end 1400 especially since the 1500x has double the L3 cache and higher clock speed than the 1400.

>>59697446
1500x and 1600 will probably be the big sellers. Why would you spend $30 more for a 1600x and still have to buy a cooler when you can just overclock a 1600?

>>59691903
Basically buy based off your use case. If your workload or daily tasks could benefit from more cores and scaleablity then go for it.
Rendering, Streaming, Encoding, Etc its perfect for. If you are simply gaming, wait for R5 if you want into ryzen.
Personally i do all three of those but i fully expect the 1600X to be a way better value, so im waiting.

>>59697446
Assuming the 1500X has that 16MB of cache, yeah.

>>59697518
Because the 1600X reaches 4.1ghz stock on 2 cores without having to manual OC.

It's not like you need one of those fucktard delidhousefirelake $90 coolers. A $30-$50 is fine. Ryzen puts out so much less heat. A 1800X puts out less heat than a 7700k, take off 2 cores and it's even less.

>>59691903
The ryzen r7 is basically a test vehicle for AMD's Naples chip.

I wish I waited for the 16-core HEDT chip :(.

>>59697601
>Because the 1600X reaches 4.1ghz stock on 2 cores without having to manual OC.
A fixed 3.9 would be better than a periodic boost to 4.1 on a couple cores.

File: desktop phenom logo.jpg (159KB, 894x588px) Image search: [Google]

159KB, 894x588px

Phenom II x4 user here. She's still kicking and handles all my demands with ease so I'll just keep her for a long while yet. Applied fresh thermal grease a while back so she's good for a long time. (and if she does kick, I've got a spare 955 stashed away)

>>59697642
Considering how the 1800x typically makes it to 4.0 on all 8 i cant imagine why the 1600x wouldn't as well. its basically the same chip with 2 cores less and half the money.

>>59695128
The fuck is this cuck, lad?

>>59693806
not only that, but it enables you to stream with a shittier connection.

>>59691903
Yes, no.

>>59697662
If they can boost to 4.1 you'll probably be able to get 4 on all cores. If the 1600s were intended to be 6 cores from the start or if they were just fucked up 7s that needed cores disabled will probably make a difference though.

>>59696991
>>59697175

The day -0 reviewers used shit ram with shit speeds on shit motherboards that would randomly fry memory due to no control over voltage.

EVERY review from that time, showing amd being good or bad is highly suspect.

>>59697225
It 100% depends on the source file, size of ram, and what you are outputting, but generally more ram trumps ram speed.

>>59697658
The 740 is probably a much bigger bottleneck than your CPU, in most cases.

>>59697642
I'm pretty sure the all core boost is like 3.8? Probably the same as the 1800x's but I forget what it's all core max is stock.

4.1 on 2 cores would actually be objectively better for games if the thread scheduler would stop hoping threads around off the boosted thread.

>>59697621
I was thinking of 12 or 16 core HEDT, but na probably better to wait until 7nm Zen. Lower power, higher clocks.
If yields are good, it may be cheaper too.

>>59697714
>If they can boost to 4.1 you'll probably be able to get 4 on all cores.
It boosts to 4.1 on 2 cores by upping voltage to them to like 1.475. lmao.
You can't all core that voltage.
So no. You might be stuck at like 3.95Ghz if you're unlucky or your mobo is being shit.

I'd rather just keep the 1600X stock. That's the reason to get it.
For manual OC, a 1600 is way better value. You'll probably get 3.9ghz on the stock cooler for well over 3/4rds of them.

>>59691984
not really, If your hobby/work would improve 10-20% by an upgrade, it's a reasonable upgrade, his cpu is on that cusp where amd is going to generally outperform it in everything.

I would say go for it due to quality of life of having 8 cores with the loads he does.

>>59697284
amd has a sizeable amount of low hanging fruit to pick that could likely increase performance 10-20% overall for programs that don't update to better utilize ryzen.

And from the process side, there is always something they could do to get clock speeds higher without cranking the voltage retardedly. I'm not expecting an 8 core 5ghz, but I am expecting a less voltage limited clock situation, where even amds 4 cores are limited to 4ghz because of voltage, something closer to intel's disable cores get faster cores, should be possible.

>>59697575
my current idle speeds are 25% cpu use, for me, even if everything I do intel could do better, that 25% load (really 25-40%, right now its just on the low end) kills peak performance

More cores would smooth any cpu hit out. for me, its like watching the amd gets 133fps here, intel gets 180, and then realizing that because of my 25% minimum constant load, you have 25% less fps off intel and that takes it to 135, but even then, amd is not impacted at all by that load,

>>59697445
Lower binned chips always are. Applies to CPUs and GPUs. Look at the success of the 750 Ti. Literally every 12-year old and his friend has one.

>>59697284

To be absolutely fair. Developers in general, have been optimizing for high end Intel hardware for nearly a decade, because of the colossal fuckup that was Bulldozer. The fact that Zen can get within 15% of a 7700k, whose whole purpose is high clocks to maximize IPC potential, despite zero optimization for the platform is pretty impressive.

And that's with buggy microcode, mobo issues, and inconsistent driver support. Since then, AMD has been releasing updates to the microcode, mobo makers are addressing issues with memory compatibility and speed; and as a result, Zen is making significant jumps up. Now, it sits within 1-3% of a 7700k.

Given the fact that Zen also basically provides near Broadwell-E levels of performance, means that development studios and applications in general now have to cater to both, as the AMD market share grows. This in turn means, better framerate in games and overall improved performance in application handling.

Intel is still ahead, no doubt about that. But the gap has narrowed enough that they are going to have to compete in the market again, instead of just die shrinking a 6700k, raising the clocks and selling the CPU as new for crazy prices.

>>59697734
Encoding doesn't take much RAM at all.

>>59697939
>Look at the success of the 750 Ti. Literally every 12-year old and his friend has one.
I just really don't get this.

Why get a GPU that only works for 1 year then suddenly doesn't work on next year's games because of gimping it'd have that only becomes apparent next year?

Historically, I think the worst you should have ever gotten was the tier below the higher end one.
A 280X, or 770, for example.

Though now days.. they are good two tiers lower. The 1060 (even if not as good as its competition for the price), and the 470 (yes, the 470 is 2 tiers lower. R9 Fury, 480, 470). That's the first time this has really been true, though.

>>59698033
Why pay so much for a GPU when most new games are shit?

>>59698757
Thats why I'm still rocking the 390x(some fag told me to buy a 970 instead of a 290x so i got memed on, I know.)

>>59697956
depends entirely on source file size, you got a 1tb 30 minute long video, that's going to be better served with more ram and faster ram then anything else.

>>59698833
No, it's decoded on the fly, and worked on in uncompressed YUV.

This seems like a good place to ask this, But does Having more Cores and Threads in a a CPU limit its per core performance because of thermal throttling?

>>59700214
no

Vidya, recording, video editing, and coding in near future.

Should I get a 6c or a 8c? Currently have a i5 4xxx

>>59697953
Yes, that's about right I feel. The microcode is not that bad: Intel's current microcode is worse in many ways.

One of the big performance drops people found with Ryzen turned out to be with the way an earlier Visual Studio compiled with A LOT of fences in certain operations, to work around an Intel erratum which would otherwise freeze the chip - since fixed in microcode via BIOS updates, but the workaround is still being used in that compiler.

Intel's replacement microcode actually ignores the resulting fences as fences, and uses them as signalling to engage the workaround. AMD didn't know about this (because it's Intel Confidential), obey the fences, and because of the strict victim cache layout, that hurts the performance of that particular app around 30%.

Compile the same app with VS2017, which doesn't have the workaround for that particular errata anymore, and that all goes away and AMD beat Intel.

Guess which one Intel chose to benchmark with. Of course they're going with FUD - if you remember the 90s, FUD was always Intel's first choice in marketing when they had a competitor at their heels, they learned it from oldschool Microsoft. The new thing here is actually having a meaningful competitor once again. Kudos to AMD for that.

If I were buying a gaming PC, I'd be weighing up the options. If I were buying it for compilation or anything else, I'd probably pick Ryzen right now, because cores really improve that, until Intel inevitably start pivoting and producing cheap 8c/16t desktop parts. Don't know if that's going to happen on their 14nm or if they'll have to wait until they finally get their 10nm process working properly.

how does it compare to my i7 920

File: 1490042646578.png (158KB, 639x349px) Image search: [Google]

158KB, 639x349px

>days since jewvidia tricks: 0
Gamewerks gimping AMD GPU's is not enough, they have moved onto gimping AMD CPU's now.

https://www.youtube.com/watch?v=0tfTZjugDeg
https://www.youtube.com/watch?v=QBf2lvfKkxA

>>59700214
It can if your cooling solution is inadequate(see Macbooks throttling to 800MHz under load), but in desktops it generally doesn't.

>>59693818
I was lucky enough to check newegg a few days ago when it was available for what seemed like a few minutes. Hang in there man, the wait will probably be worth it

File: ryzen-photoshop.jpg (79KB, 701x479px) Image search: [Google]

79KB, 701x479px

>>59700808
>anything else

Cores really don't improve anything else though. They are for very specialized highly parallel tasks like encoding and rendering.

>>59693818
>tfw got a taichi but waiting on AM4 mounting kit

File: image.jpg (43KB, 630x384px) Image search: [Google]

43KB, 630x384px

>>59701780
>let's cherry-pick programs which still feature little to no parallelization even though some parallelization is far from "very specialized"

i have a 4k monitor and only a gtx 1080
the cpu bottleneck will not hit me

>>59703800
oops, I checked the box and it's 3.840k aptually

I don't even care, Ryzen is pretty good, and there is a lot of potential. I bought Ryzen, and an rx480, it just feels right, yes, I am an AMD fanboy, I've never built an intel platform. so far, it has handled everything I've been wanting to play, on my 1440 60hz IPS panel.

I am just glad they made something that can compete again, so I will take a slight performance hit compared to intel right now, just so they can have some competition. Someone needs to do it.

Intel fans should be really happy about Ryzen, Intel has been really lazy lately.

>>59694251
>Look Mom, I posted it again!

>>59691903
gayest name ever

>>59698757
Well this is a good point now. But less so 6 years ago.

>>59701308
I don't think Nvidia is gimping Ryzen on purpose.
If they were, it wouldn't be through stuff you could just change in settings. It's just defaults that are messed up on Ryzen CPUs.

>>59701780
>very specialized
>cherry picks what's actually very specialized to shill for kabylake
lmao.

That result is close enough that you're still not losing hardly anything getting a Ryzen 7 instead for productive use in that cherry picked example, while you get massive gains elsewhere.

You're either an idiot or a blind shill if you don't get that.