Which idiot thought it would be a good idea to market Ryzen as

I don't understand why you make this thread
What gain do you get from this?

>>59345090
I don't understand why you make this post
What gain do you get from this?

>>59345090
>gain
everything isn't a coldly calculated jewish scheme to maximize personal gain
some things are more important than that

The inconsistency between games , MT / ST benchmarks really points out some software issues.
How can it do as well as a 6900k in production and then shit the bed in gaming?

>>59345684
Windows 10 RTC disliking 0.25x multipliers, causing timing issues,
Software not reading L3 properly (thinking each core has 8MB of L3, rather than 2MB/core),
Latency within a CCX being regular, but across CCX boundaries having limited bandwidth,
Static partitioning methods being used shows performance gains when SMT is disabled,
Ryzen showing performance gains with faster memory, more so than expected,
Gaming Performance, particularly towards 240 Hz gaming, is being questioned,
Microsoft’s scheduler not understanding the different CCX core-to-core latencies,
Windows not scheduling threads within a CCX before moving onto the next CCX,
Some motherboards having difficulty with DRAM compatibility,
Performance related EFIs being highly regularly near weekly since two weeks before launch.

File: buggy_radv.png (35KB, 915x489px) Image search: [Google]

35KB, 915x489px

>>59345072
Open source AMD Vulkan RADV drivers are buggy.

>>59345684
AMD sacrificed gaming performance for MT and power consumption, this whole gaming issue stems from software thinking the CPU has two 4 core CPUs instead of a CPU with two 4 core clusters.

Not this thread again.

>>59345778
Even if the cpu is only using 4 cores fkr games, it doesn't explain how it can be worse than an i3.

>>59345886
Because the scheduler doesn't keep the cores on one cluster, but throws them around

>>59345899
>threads*

File: l4700ozjqg.jpg (11KB, 132x147px) Image search: [Google]

11KB, 132x147px

>>59345899
Is zen a dual quad core processor?

>>59345957
No, but it looks similar to Windows.

>>59345072
amd released a slide saying ryzen is good also on gaming

shills thinks amd marketed it as a gaming cpu

wew lad

no suprise here

>>59345701
actually windows thinks that the cpu has 138mb of l3 resulting on tanking the system ram with random mumbo jumbo also

half of those problems are related to the scheldurer patching

>>59345957
>>59345960
yep
https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet

low performance happen when 2 CCX trying to communicate

>>59345072
Remember, there are more ryzens coming..

>>59345684
>The inconsistency between games , MT / ST benchmarks really points out some software issues.

Anybody that says this has no idea how game development works. Game development is all about smoke and mirrors: the entire thing is about optimizations, because the real world is too complex for our hardware to try to recreate.

What this means is that there will always be this kind of inconsistency in video games. Video game developers will always be able to use the performance given to them. For them it's a question of how much optimization they need to do and which kinds of features they need to cut. Performance (rendering) is one the reason why games like Battlefield are capped at a certain number of players per server.

>>59346404
HOLY FUCK

The fact that Windows 10 is balancing the 8 thread load specifically between matching logical core pairs indicates that the operating system is perfectly aware of the processor topology and is selecting distinct cores first to complete the work.

microsoft already publicly addressed that they will release a fix and pcper is saying its all good nothing substantial to see here

fucking hell that is beyond any stupidity

>>59346451
It's Microsoft. They seem to think that it's a good idea for Windows 10 to have separate Windows 10 style menus for things such as Add/Remove Programs and to also have the old Windows 7 and older style Add/Remove Programs. Oh, and these are not in sync nor is this the only redundancy in software in Windows 10.

Microsoft also seems to think that running code for network adapter discovery should be run on the UI thread of the network and sharing center.

Microsoft also thinks that in Windows 10 that if you have network paths set up to query them each time you open a file saving dialog. While the request doesn't run on the UI thread it seems to require a response or timeout before allowing you to close that dialog. This means that if you have a disconnected network path your file saving dialogs might take longer to close than you would like to.

File: 1336501025.png (11KB, 800x401px) Image search: [Google]

11KB, 800x401px

>>59345072
soon.

>>59346404
>>59346451
Why not test with 4+0, disabling one CCX? Their test about SMT is irrelevant.

>muh scheduler
There is a reason the scheduler swap threads on cores. It's by design to even the load out on all the cores and have better temps.

>>59346554
>Why not test with 4+0, disabling one CCX?

it's same people that said 1060 will be 60% stronger than 480. what do you think?

>>59345276
Like what? Wasting away the prime of your life on an imageboard?

>>59346554
because you cant select what ccx to disable..and if they are on the other part of the infinity wiring you will still get stupid results..

if the ccx's are on the same die then yes you will get a massive upgrade mainly because of the massive L3(there is a reason as to why both amd and intel have an insane price to the cpus with massive amount of l3 instead on those with high clocks)
>>59346578
that is how intel do it not amd...the threads on amd taps into l1 l2 and l3 cache

>>59346601
can scheduler use only one ccx when it gets patched for 4core loads? like dual xeons do it?
that giant l3 for 4 cores going to be priceless on less threaded loads if it's managed right?

>>59345072
Only kids from /v/ thinks an 8 core 16 thread CPU is purely for gaming, AMD marketed it as being capable of gaming, especially when doing things like streaming at the same time.

>>59346648
6900K is very good chip for gaming if it was 400 like ryzen
everyone, except most shilled reviewers, complained about that raw performance is there but it doesn't translate into gaming

>>59345072
K̶̵̢̩̺̲͕̖͖̰͖̫ͬͩͯ̆͆ͧ͆̽̃̓ͧ͒͘Į̴̨̹̝͎̙͎͎̞͙͉̟̻̲̯̝̗͉́̒͋̒̏́̃͟L̶̡̬̻̤̿ͨͧ̃ͯ͌͆̆̚L̡̧̠͍͚͕͙͔͉͕̖̟̮̗̙̺̭̖͕͚̎ͫ͋͋̿͑ͤ͋̾͟ͅ ̢̼͚̭͉̪̲̗̣͎͇͉̩̠̘̩̄̍̀ͬ̔ͯͩ̔͞ͅY̧̛̟̺̖͉̞̬͉ͫ͑ͣ̌͛͒̀̾̑͌͡Ó̢̡͎̹͉̳̤̞̎̀̋͒̍̏͗̅̒̐͜U̩̫͎̮̳̥͚͕͈̙̩̞̯̞̠͖͗̓̆̑̂̌̀̋͛̀̚͢͡͡R̡͔̺̼̅̒̍̽̽̚͢͜͞S̶͔̠̗͖̙̘͎͍̥̟̼̣͙̞̣̳̞ͨͪͤͦ͑̄̄̑͛͗̚͝͞E̸̝̰̜̜͖̬̣̘̟͍̽ͮ͊̿̎̐͛͛͗͋̇̅̋͆̇̊͛̓Ļ̸̙͚̦͉̹͈̙̤̩̻̟̩̹̔ͣ͊̀͗͆̎̊ͫ̀F̡͚̭̞̼̟̼̜́͆ͯ̓̓͠ ̷̲͇͈͈̜̟͈͚͙̰̿͗ͫ͐ͩ͌ͪ̑ͨ͆̔̊̚͘͟Ř̵̵̡̫̰̹̬̪̮͉̹̹̤̱̯͍̦̝̾̑͐̋̄͆ͩ̑̽͊͋ͨ̇͢ͅE̷̩̗̲͇̘̪͓͓͈̹͓̞̙̯̺̲ͤ̍́̐ͧ̑̎͜T̸̛̪͈̣̭̃ͦ͆̓ͭͩͯ̂̀ͦͪ̍̓͠A̛̤̼̥̟͖͕͓̺̻̟̟̰̝̫̬͑ͩ̎͂ͫͤ̑̐͛̔ͨͮͭR̷̠͙͈͍͕͈̭̘̺̣͐̓̾̇͐͌̄͗̈̆̓̀D̢͔̗̺̬͔͖̑́̋ͭͭͬ̉̏ͤͫ͢͞Ȩ̢̩̣̤̼̲̘̣̝͇͂̇̈ͣ̅̓ͫ̂͊͌̓̽͗͜͞͡D̡̩̮͉̦͓̞̦̝͖̆ͤ͑̅͜͢͟ͅ ̡̘̬̜̜̭̩̦̘̻̓̒͐͊̿̉̎ͫ̍ͭ͋ͭͤ͝S̸̛̙̤͚͖̳̹̳͕̜͌̌͛̐Ḧ̶̬̣̝̖̙̥͙͔̝̲͇̲͇̱̆ͧ̆̂ͬ̾ͪ̀̌̊ͪͣͬ̚͘͢ͅI̧̡̬͈̺̞̼̖̫͍͉̲̪̜̬̰͔͕͉̳ͫ̏̌̄ͭͮͭ͋ͬ̏̈͒̾̿͗ͣ̎́͘L̨͚̻̣̞͙̽ͩ̿̑̏ͦ͋ͧ̈́͋̈́͑̾̿͌͘͜͡Ļ̡̬͖̝̻͙̪̫̠̘ͮ̾ͦ͋ͧ̒͐̍́̓̄ͮͬ̉ͦ̄̀̚͢

File: pcper.png (48KB, 984x297px) Image search: [Google]

48KB, 984x297px

>>59346601
>because you cant select what ccx to disable
You can? Supposedly they will test this too

>if the ccx's are on the same die..
Afaik Ryzen l3 is not shared between CCX, it's a victim cache

>>59346680
no you can select which core to disable but you have no idea on what ccx they are

its shared among the whole die tho obviously its 8mb per die 4 per ccx

also i wonder if 7700k is so good at gayming can it do this?
https://www.youtube.com/watch?v=Njm0MBOwFTM&

running 4 unreal 2 games and 2 unreal 3 ones

gotta love to see it crash and burn

>>59346704
>8mb per die 4 per ccx
Really?

>>59346680
He is blaming windows 10 when it works exactly as intended. AMD fucked up the memory controller design.

>>59346704
>running 4 unreal 2 games and 2 unreal 3 ones

This kills the 7700k

File: scaling.jpg (105KB, 650x382px) Image search: [Google]

105KB, 650x382px

>>59346704
>no you can select which core to disable but you have no idea on what ccx they are
What makes you say that? There are BIOS settings for that, 4+0 should disable one CCX.

>>59346813
Jesus christ, this is pathetic, how am I supposed to play 4 games at the same time? INTEL BTFO

>>59346824
>he can't play multiple games
Is your brain single-core or what?

>>59346704
>>59346750
> shared among the whole die 8mb per die 4 per ccx
16 mb die, 8 per CCX

https://www.techpowerup.com/231268/amds-ryzen-cache-analyzed-improvements-improveable-ccx-compromises

"The fact that this memory is "mostly exclusive" - which means that other information may be stored on it that's not of immediate use to the task at hand - can be responsible for some memory accesses on its own. Since the L3 cache is essentially a victim cache, meaning that it is filled with the information that isn't able to fit onto the chips' L1 or L2 cache levels, this would mean that each CCX can only access up to 8 MB of L3 cache if any given workload uses no more than 4 cores from a given CCX. However, even if we were to distribute workload in-between two different cores from each CCX, so as to be able to access the entirety of the 1800X's 16 MB cache. we'd still be somewhat constrained by the inter-CCX bandwidth"

Disabling one CCX should leave only 8MB L3 cache

>>59346821
but you dont know which core can overclock better than the others

this is what ryzen master oc app does

BUT it doesnt say on which ccx it sits lol

>>59346824
you know how many mmo farmers will love it?

i was able to open 16 instances of lineage2 on a 631 and the money i got from that game each month was barely enough for my electrycity bill..

with this one i will probably open more than 30-40(assuming i have more than 16gb) and the cpu wont be even near to its limits..

>>59346891
Nobody is talking about overclocking. You CAN disable one CCX eliminating inter-CCX latency until windows scheduler is fixed, this ought to be tested.

>>59346917
Get a real job.

>>59346925
>until windows scheduler is fixed
What's broken?

File: ping-amd.png (159KB, 602x477px) Image search: [Google]

159KB, 602x477px

>>59346940
Moving threads between cores in different CCXes has a significant latency penalty. PCPer test showed that but they didn't test with disabling one CCX

>>59346968
How is it windows 10's fault? The scheduler moves threads between cores for a reason you know.

>>59346976
Scheduler should prefer moving to a core in the same CCX rather than doing an inter-CCX move.

>>59347004
Why was it never a problem before ryzen?

File: bulldozer.png (111KB, 550x194px) Image search: [Google]

111KB, 550x194px

>>59347016
Because architecture was unified? Bulldozer modules shared the same l3 cache rather than having it split between groups of 4 cores.

>>59347016
because windows 10 has special scheduler, with special bells and whistles thus w7 got 20% perf bump

you see those numbers? 40ns ->140ns it's x3.5 slower than it should be
heck I can't believe this thing even works as good as it does

File: When you find out Rysen is pulling the same shit as Bulldozer.jpg (97KB, 435x435px) Image search: [Google]

97KB, 435x435px

>>59347016
Because no one else was so god damn stupid to build something like this. This is the exact same reason Bulldozer failed. 2 Cores per core module. It was a terrible idea then and it's a terrible idea now!

>>59345072
Fully agreed
The Ryzen gaming slides leading up to the R7 launch has basically damaged their rep already.
What if they had waited and been all like "AND NOW FOR THE GAYMEN SHITE" when releasing the R5

>>59345072
Oh look AMD fails... Again.... For the nth time...

File: 1488634431590.png (609KB, 854x640px) Image search: [Google]

609KB, 854x640px

>B-B-BUT AYYMEEDDEE WILL COME AROUND WITH R5 JUST WAIT™

>>59347062
Yeah, the scheduler is making it so the charge is balanced over the cores for less temperature and more efficient consumption. I don't understand what's the problem exactly.

>>59347054
Why didn't they make a unified architecture?

>>59347149
Why are you still living in your mother's basement?

>>59346917

Lineage 2 is dead, thanks for bringing it up ;_;.

You don't happen to know of any good private servers that's not 100% BR or Greek?

>>59347149
Modular design, smaller cores, power savings optimized. You can scale this to 24 core server CPUs to 4 core apus.

>>59347081
the 6 and 4 core won't be better cpu wise, just price, they will both throttle at the same areas for the same reasons, but voltage needed to push it higher would damage the cpu.

amd has 2 problems

1) bios side, all venders were given 3 weeks to make something as that's when they received final versions.

2) windows 8 and 10 throwing shit around at random thinking threads are cores, then moving things between the two core complexes for no real reason.

Once these are sorted out, the shit that happens in ops benchmark will go away. shit won't have the brute force single core as a 5ghz 7700, but it will give stock a run for its money with xfr enabled.

>>59346794
yea... never say windows 10 works... it functionally dysfunctions and people confuse that with working, but its not working.

>>59346575
there a way to turn that crap off... don't really want it going from core to core, seems like it would fuck with any processors performance.

>>59345072
Lol AMD is so fucking retarded

I can't wait until that piece of shit company is dead in the ground

>>59347234
well no there are some good greek ones but thats it (avoid ofc bnb at all costs)

>>59346554
Resource sharing between threads in the same core is also different from Intel's HT, and thus require code specific to Ryzen's SMT.

>>59345072
>workstation cpu
>not a single motherboard with ECC-support available
git gud

>>59347622
It's not openly advertised, but it works. Or it will work when BIOSes are fixed.

>>59345684
The CPU's/chipset has multiple issues. They are working on it.

File: 65681803[1].jpg (106KB, 298x403px) Image search: [Google]

106KB, 298x403px

>>59347635
>Just Wait.

>>59347622
it works, wendel says so, it is so

>>59347654
See
>>59347657

ASRock is the only manufacturer advertising ECC support, but it's not working yet AFAIK.

>>59347657
>AMD shill

>>59347689
Not a single manufacturer is advertsing the RAM being used in ECC-mode.

If anything, they're advertising to support ECC-modules (duh, pretty much every mobo does), but in non-ECC-mode.

>>59347071
I know you get moist by parroting that, but all of Bulldozer's problems are orthogonal to the modular design.

>>59347694
>knows his way around hardware and software
>AMDshill

I'm really not surprised.

>>59347071
You're clinically retarded.

AMD shills are so pathetic, constantly making excuses and saying "just wait"

>>59347724
http://www.asrock.com/mb/AMD/X370%20Taichi/index.asp#Specification

>- Supports DDR4 3200+(OC)/2933(OC)/2667/2400/2133 ECC and non-ECC, un-buffered memory*

How's being a retarded shill treating you?

>>59347865
Taichi and Aorus are the 2 best mobos around at the moment.

Ironically the ASUS ones are pure junk

>>59347898
MSI is also shit

>>59347865
Thanks for confirming my statement.
The mobo supports ECC-RAM, but doesn't use it in ECC-mode.

>>59347960
https://www.reddit.com/r/Amd/comments/5x4hxu/we_are_amd_creators_of_athlon_radeon_and_other/def6vs2/

>>59347960
>Supports DDR4 3200+(OC)/2933(OC)/2667/2400/2133 ECC and non-ECC, un-buffered memory
Which part did you not comprehend?

>>59347960
Every desktop motherboard in existence works with unbuffered ECC ram. They just ignore it.

>>59347982
There's no sign of an ECC-setting in the BIOS, so it obviously doesn't run ECC-RAM-modules in ECC-mode.
It's rather easy to understand, anon.

Regarding forcing ECC in leenox:
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=044e7a414be9ba20826e5fd482214686193fe7b6

Another
Massive
Dissapointment

>>59345701
Eh. You're along the general lines. This is, after all, a new chip.

If you saw the inside of a GPU driver - any GPU driver - you'd shit the bed about how much it cheats and how many workarounds it has.

Wait until you try and use Intel's 3D X-Point - without deep software support that will require extensive OS redesigns, you won't get the performance out of that which you could otherwise, which is why that's going to be enterprise-only for special applications for the next 5-10 years.

>>59346451
The Windows executive scheduler doesn't yet know about the latency between CCXes or how to remap parts of the L3 to do zero-copy communication between them (yes, it can - this is one potential use for CLZERO but requires special support) - so everything goes to main RAM and back. It'll be addressed in the Creators Update.

Linux has some support, but only if you're running development kernels.

Intel's upcoming MCMs will exhibit similar behaviour, but they're not planning to launch those in the consumer space at the current time.

>>59347960
>>59348033

Wendel tested and said ECC mode works in Linux.

https://youtu.be/G_6rs9cBzvE?t=524

>>59345072
>It's clearly a decent workstation for the price
Does it have ECC RAM support?

>>59348050
Yes.

>>59346501
>more expensive
>lower min and average FPS than the 7700k
Lol

>>59348050
CPU does, boards don't.

>>59348045
The change to disable ECC-forces comes in 4.10, does he say which kernel he used?

>>59348084
(you).

>>59348093
>>59348099
He used 4.11-rc for testing actually in his asrock review
https://youtu.be/VjGWqTFultI?t=563

ecc mention at the start of video
https://youtu.be/VjGWqTFultI?t=85

>>59348035
>It'll be addressed in the Creators Update.
Source?

>>59348173
Google it.

>>59345090
Are you not doing the same?

>>59348184
Didn't find anything except some hope windows "game mode" will fix it.
I doubt MS can patch a workaround for Ryzen in their scheduler in the next 2 weeks.

>>59348210
Dude it's already patched on Linux. Patching scheduler is not that hard.

>>59348050
Asrock Taichi does support ECC tested in linux
>>59348167

https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet
>In fact, though we are waiting for official comments we can attribute from AMD on the matter, I have been told from high knowledge individuals inside the company that even AMD does not believe the Windows 10 scheduler has anything at all to do with the problems they are investigating on gaming performance.
Oops.

>>59345072
I don't understand this benchmark. You can get 60 FPS out of DOTA 2 on a potato attached to a wallsocket.

>>59346404
its bulldozer 2.0

File: THANK YOU BASED JIM SHITWRECKER KELLER.jpg (193KB, 1280x1024px) Image search: [Google]

193KB, 1280x1024px

>>59348227
>pic
noice.

>>59348234
ANOTHER
MASSIVE
DISAPPOINTMENT

>>59346404
>In fact, though we are waiting for official comments we can attribute from AMD on the matter, I have been told from high knowledge individuals inside the company that even AMD does not believe the Windows 10 scheduler has anything at all to do with the problems they are investigating on gaming performance.

>we are waiting for official comments we can attribute from AMD on the matter
>I have been told from high knowledge individuals inside the company

Good God, why are people falling for this?

File: 1458521351619.png (141KB, 750x829px) Image search: [Google]

141KB, 750x829px

>>59348223

People non ironically believe MS will change their scheduler for one defective architecture?

>>59348314
They did it for IA64 so yeah, I don't see why not.

Well, good thing linux fixed it. I might finally switch.

Anybody has comparative benchmark of BF1 between windows and linux?

File: amdgpu.png (38KB, 739x507px) Image search: [Google]

38KB, 739x507px

>>59348240
>>59348297
Because it's a Vulkan GPU driver bug. Even with shitty AMD OpenGL drivers it gets 80 fps. I wish there was an Nvidia + ryzen linux test

>>59348314
~%5 gains across the board, maybe. Nothing wrong with Ryzen as is though.

>>59348371
>Nothing wrong with Ryzen as is though.
Obviously there is. Why is it shared cache if it's based on double CCX that cripples perf when some cores take datas from some part of the cache?

>>59348388
Architectural decision to favor scalability first. AMD is betting on its 32-core Naples for servers

>>59348314
Neck yourself, shill.

>>59348234
>hey, I posted it again

File: dog kills baby.webm (1MB, 400x400px) Image search: [Google]

1MB, 400x400px

>>59346600
>Like what? Wasting away the prime of your life on an imageboard?
https://www.youtube.com/watch?v=P22gcb4YHso

>>59348431
>architecture decision
Alright, so it's by design.

>>59348487
And what's wrong with it?

>>59348503
Cache has latency in gaming related scenario and it cripples the perfs?

>>59348503
Nothing, he's mad because it's not a gaymer CPU

>>59348518
>gaymen
So, who cares? 4/8 will be 1 CCX for GAYMURR faggots.

>>59348527
Will it be ? Don't get me wrong, I did believe and I wanted it to be a success. Any source on the 4/8 with 1 CCX ?

>>59348543
The source will look like a piece of rope. For you.

>>59348543
>Any source

If a CCX is 4/8 and the Ryzen 3 is 4/8 (or 4/4), how many CCXs do you think it will have?

>>59348527
>So, who cares?
Everymotherfuckingbody on this board since ryzen release?

>>59348371
>~%5 gains across the board
it's more, 40ns vs 140ns
normal cache vs overwritten cache every other op
not to mention obvious - smt is not working

windows knows it's NUMA, it doesn't know which ccx is which
worst case it's going to be linux/w7 performance which is already 15-20% more

>>59348563
You mean /v/ crossboarders? These are not humans, but animals.

aren't there benchmarks with cpu-affinity set to stay within a CCX?

>>59348580
There are some (albeit, questionable) 4/8 1 CCX benches.

>>59348543
>>59348580
http://www.zolkorn.com/reviews/amd-ryzen-7-1800x-vs-intel-core-i7-7700k-mhz-by-mhz-core-by-core/3/
Only this, and that source is still shaky. 7700k still have IPC and frequency advantage

>>59348562
>how many CCXs do you think it will have?
I don't know. Why would you think they won't have double CCX? maybe it will be a 2+2 split. I just wanted some source.

>>59348580
kind of - not enough, they put it up there to 6700K levels of performance in games with 4c

> all this uncertainty about ECC

Is a proper server CPU from AMD of similar capacity more expensive?

>>59348594
What the fuck, rajeet.

File: 1485065188739.jpg (23KB, 800x320px) Image search: [Google]

23KB, 800x320px

>>59348598
there is only this that leaked from closed part of conference

>>59348619
I don't see where it says 4/8 will be single CCX.

>>59348608
they probably found texting many times which core belongs to which ccx and grouped them right

>>59348631
that's the point - nobody knows

but isn't it easier just saw r7 in half? they work perfectly well in halves
if anything I think making 6c from it is harder
also imagine if 4c gets 8mb of cache

>>59348565
>worst case it's going to be linux/w7 performance which is already 15-20% more
https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-8#post-38775732

For minimums and smoother frametiming sure, it will be much helpful, but it's like 8% increase for average and that's for one game. When many tasks and games are averaged it might not be the magic people wanted, still a good increase. Bulldozer patch was much like the same, in the end it didn't reverse roles

>>59345072
They meant the Ryzen 5.

File: AMD-Ryzen-6-Core-Processors.png (182KB, 831x704px) Image search: [Google]

182KB, 831x704px

>>59348598
>Why would you think they won't have double CCX?

Cost. Selling a 8-core die with 2 cores disabled isn't much of a big deal, but with 4 cores it becomes hard to recoup the investment.

>maybe it will be a 2+2 split

See above.

>I just wanted some source.

AMD slides show a schematic Ryzen 5 die with two cores disabled, but they don't show one for the Ryzen 3.

>>59348659
>that's the point - nobody knows
Then why did he tell me this. What's the point in lying? I just want a cheap good CPU, I want them to be great.

>>59348662
w7 just doesn't shuffle threads as aggressively, it still shuffles them as far as I know

>>59348671
I see. Thanks for the detail, I hope you're right.

>>59348671
damn, a 6-core ryzen gets a lot of cache per core.

>>59348683
At least that explains the minframe difference.

>>59348371
>Because it's a Vulkan GPU driver bug
Why would anyone benchmark a buggy driver? What is even the point of this shitty thread?

File: 1457955653376.jpg (93KB, 978x546px) Image search: [Google]

93KB, 978x546px

>>59348671
I looked into what you said. I found this article in my country

http://www.comptoir-hardware.com/actus/processeurs/32270-quatre-coeurs-zen-creent-un-cpu-complex-ccx.html

>4 cores make one CCX unit
>AMD assures that with this design they'll be able to make 4 cores CPU with an APU taking the place of the second CCX.
So I'd say you're right and ryzen 3 will be single CCX.

>>59348782
Michael is a known Intel shill.

>>59348786
>4 cores APU with an GPU
Fixed, my mistake.

>>59348782
Benching a pure AMD FOSS system. It doesn't make sense for performance though

>>59348794
t. AMD shill

Is ryzen 3 the real deal for destitute gamers like me ?

>>59350560
It will be when it's out. Right now, the best bet for turbopoorfags is used prebuilts with i3-2130 or i5-2400 with a refurbished gt 750 ti stuffed in there.

>>59350484
Stay mad, Michael.

File: amd btfo.png (17KB, 600x600px) Image search: [Google]

17KB, 600x600px

AMD BTFO

Can I get a non-troll response

The issues AMD is having with Ryzen right now, can they be fixed/patched in the months ahead? And if so, can Ryzen get close to 7700k performance in 1080p? Or will that never happen?

I'm on an aging 2500k and I don't want to buy the 'Jew' again.

>>59351578
There is definitely some kind of software issue negatively affecting performance at least on Windows 10. Windows 7 apparently has like a 10-15% performance advantage oddly enough. Motherboards also shipped with buggy BIOS/UEFI which introduced their own kinds of issues. These problems can be resolved but no one knows exactly how much the fixes will improve performance just yet.

>>59351578
>Or will that never happen?
it will get to 6900k performance in gaming if at least half theories aobut performance are right, it will never catch up to 7700K in old games
6900k beats 7700K in some modern games

>>59351578
Months? You'll see some improvement wrt memory and latency issues. Games? In 2-3 years when games use more than 4 cores.

>>59351657
>In 2-3 years when games use more than 4 cores.
there is enough evidence that 6900K is superior gaming cpu right now
if it was 400 i'd buy it last year in heartbeat

>>59351578
do you have any real reason not to buy intel besides meme reasons?
$299 for i7 7700k at Microcenter if you live near one. $309 on eBay if you dont.

So, where are Windows 7 benchmarks of several games?

How hard is it to get 7 running on 7 properly?

>>59351713
>$309 on eBay if you dont.
I heard enough stories not to. I wouldn't recommend lottery.

File: 1463562428373.jpg (35KB, 439x290px) Image search: [Google]

35KB, 439x290px

>>59345276
nice try

>>59351718
you can make it run, but drivers don't work
so it's simply prove of concept for now

>>59351718
>>59351748
Windows 7 drivers are unofficially supported, it's not that hard to set up, essentially same as installing w7 on skylake

http://www.anandtech.com/show/11182/how-to-get-ryzen-working-on-windows-7-x64

>>59351713
I have a sour experience with my current Intel product. It throttles down and stutters in games. Maybe I got a bad binned 2500k - who knows. The thing that also is making me apprehensive about buying a 7700k, is that I read that Kaby Lake runs hot. I don't need a space heater. I live in a desert. Finally, I'm autistic and like to have 30 tabs open, a twitch/podcast going, and vidya open. I think an 8 core is more up my alley.

>>59351899
https://www.youtube.com/watch?v=Njm0MBOwFTM

here is your answer then.

>>59351899
r7 1700 runs quite cool due to 65w tdp and has a good stock cooler

>>59351988
Holy fuck. Is it at stock?

>>59351988
>>59351997
Wow. It looks pretty good. My 2500k would never be able to do that. Nor would a 7700k either, right?

>>59352194
The 7700K would stutter like hell

>>59352215
o _ o

This is the power of 8 cores! I think I'll grab the 1700, do whatever OC I can, pray for updates, and then maybe see about upgrading to Zen+, which I'm sure will be more refined.

>>59351899
> Finally, I'm autistic and like to have 30 tabs open, a twitch/podcast going, and vidya open
This is not autism, this is optimal usage of available memory. We are not in 2GB RAM world anymore, you don't need to close one application to run another, you can have it all at once.

File: forhonor4CPU.png (104KB, 582x1048px) Image search: [Google]

104KB, 582x1048px

>>59352391

>you don't need to close one application to run another, you can have it all at once.

Only if you buy a x99 chip goy.

>>59352391
Honestly i would like to see a someone run benchmarks with a 1080p 60fps youtube video/twitch stream running in the background.
It's pretty common to have media running on the second monitor while you play games.

>games
>gnu+linux
>valve
choose one

>>59352897
???