|
GRINDCORE MEGGIDO posted:Isn't that before and after the same as disabling HT? I suspect Game Mode is doing something to limit background process CPU activity or bump process priority for things it identifies as games in addition to avoiding the HT cores. There is a lot of CPU activity occurring on logical cores 8 and 9 on that first screenshot which makes me thing there was something bad with the test methodology, though it does fairly accurately represent a real world scenario. That's going to cause a lot of latency problems with multithreaded loads as HT pipeline optimizations do their thing. The HT cores seem to be impacting performance primarily on cores 0,1,4,5 which is probably an artifact of the cache layout of the processor that MS also optimized for.
|
# ? Mar 20, 2017 22:44 |
|
|
# ? May 30, 2024 13:14 |
|
Ha ha, gently caress everyone who did testing in the past week. RIP Linus, since they just came out with a video simulating R3 and R5 perf by disabling cores and dialing in clocks to match the leaks. They *really* must be out of ideas if THAT'S what they've been doing. Still annoyed I didn't even get a chance to offer up my CNC portfolio and video ideas, though. And all because I need one of those impact statement things to get a work permit. Darn you, Dennis, and your visa shenanigans!
|
# ? Mar 20, 2017 22:50 |
|
Has anyone tested 2x2 vs 4x0 performance?
|
# ? Mar 20, 2017 22:54 |
|
SwissArmyDruid posted:Ha ha, gently caress everyone who did testing in the past week.
|
# ? Mar 20, 2017 22:56 |
|
Alereon posted:Anandtech proven right once again This is a joking reference to them being too short-staffed to get their Ryzen gaming testing done on time. It's a shame that Anandtech is in such a bad place since Anand left. No other publication has really stepped up to fill the void.
|
# ? Mar 20, 2017 23:00 |
|
Rastor posted:I think people were interpreting "no problems with the scheduler" to mean "Windows 10 performance is as expected" and that wasn't the correct interpretation. Yep, in engineering, it's a different interpretation. For example, if you overloaded a bridge and then it broke, the designers would say that it's "working as intended". That is, the bridge was design to support a certain weight, and if it goes beyond that, failure is expected. In this case, Windows was working fine because that's how it was designed. There was no "bug" with the design or the CPU. The design just needed to be changed to take into account Ryzen's differing architecture.
|
# ? Mar 20, 2017 23:13 |
I think you should probably wait for some better info than two screenshots from some blogger account with 300 followers before issuing sweeping conclusions and condemnations
|
|
# ? Mar 20, 2017 23:18 |
|
Pryor on Fire posted:I think you should probably wait for some better info than two screenshots from some blogger account with 300 followers before issuing sweeping conclusions and condemnations Considering Bits and Chips in subsequent tweets said that UT3 was the only game that they noticed any change in, I would go further and say that the conclusions drawn as a direct result of this is almost certainly inaccurate.
|
# ? Mar 20, 2017 23:33 |
|
I might agree... except that Wendell got some weird results along the same lines too: https://twitter.com/tekwendell/status/843897707135668224 https://twitter.com/tekwendell/status/843897889277526017 I don't know of any world in which you absorb a 300 MHz underclock and come out with a *higher* benchmarking score, barring some kind of freak occurrance.
|
# ? Mar 20, 2017 23:43 |
|
Ragingsheep posted:Has anyone tested 2x2 vs 4x0 performance? http://www.hardware.fr/articles/956-24/retour-sous-systeme-memoire-suite.html
|
# ? Mar 21, 2017 00:03 |
|
SwissArmyDruid posted:I might agree... except that Wendell got some weird results along the same lines too: Looks like he said "enough of this automatic overclocking crap" and futzed with the p-states to get it up to 4.1 GHz on all cores. Wendell is my favourite internet computer wizard.
|
# ? Mar 21, 2017 00:42 |
|
Kazinsal posted:Looks like he said "enough of this automatic overclocking crap" and futzed with the p-states to get it up to 4.1 GHz on all cores. I've done some benchmarking on Ivy bridge Xeons which we bought to make up for some terrible query writing, and disabling power management boosted performance by 25%. Even with the hardware monitor telling you that all of your cores are running at boost clocks, if you let it the system clock itself it will hamstring your performance by downclocking cores for milliseconds at a time. I don't know if Intel has improved on this with newer processors, but it isn't at all surprising to see it hit AMD too.
|
# ? Mar 21, 2017 00:51 |
|
So amd unfucked their stuff?
|
# ? Mar 21, 2017 02:24 |
|
Also spotted in the new Dota 2 patch:Valve posted:Improved threading configuration for AMD Ryzen processors. the ride never ends reviewers, you're re-benchmarking ryzen forever
|
# ? Mar 21, 2017 02:50 |
|
Blorange posted:I've done some benchmarking on Ivy bridge Xeons which we bought to make up for some terrible query writing, and disabling power management boosted performance by 25%. Even with the hardware monitor telling you that all of your cores are running at boost clocks, if you let it the system clock itself it will hamstring your performance by downclocking cores for milliseconds at a time. I don't know if Intel has improved on this with newer processors, but it isn't at all surprising to see it hit AMD too. You generally need a very poorly written application where its executing in incredibly small increments with pauses to have that kind of performance penalty. Even back on the first processors started supporting speedstep the time to transition between states was in the nanoseconds and since then intel has done more work to optimize against "fluttery" workloads that have this kind of problem. Its not impossible, but it is uncommon and an edge case where most people are better off leaving the power management to the OS or whatever OEM power management chip they put on the board. Turning power management off will make things idle at a higher temp and impair access to P-states when it needs to ramp up for a sustained burst workload.
|
# ? Mar 21, 2017 03:20 |
|
wargames posted:So amd unfucked their stuff? at this point between Polaris and Ryzen we should just assume that AMD hardware is going to launch like the average AAA video game: on fire at release day, a great buy 3-6 months after
|
# ? Mar 21, 2017 03:37 |
|
wargames posted:So amd unfucked their stuff? For some games yes, for others no.
|
# ? Mar 21, 2017 03:41 |
|
fishmech posted:For some games yes, for others no. Which games though? There are newer games that scale very well with the number of CPU cores/threads a system has while other games only scale well with CPU clock speed regardless of the core/thread count because they are poorly optimized piles of poo poo. spasticColon fucked around with this message at 05:40 on Mar 21, 2017 |
# ? Mar 21, 2017 05:35 |
|
Generally speaking: Any console games whose development began after Ryzen's launch, and also receives a PC port. Remember that this... all of this, is part of a greater AMD strategy. As AMD powers both the XBone and the PS4, the Scorpio and the PS Pro, and probably has a greater than 50% chance to power their successors afterwards. (on the Microsoft side, it is almost 100%, there's almost zero way they don't continue with x86 as an architecture, Sony can always decide they want to do something different, but x86 offers them some very tangible benefits they are unlikely to relinquish after one console generation, and Nintendo.... well, Nintendo is just gonna keep doing whatever the gently caress Nintendo does.) That means any games whose development begin on AMD console hardware, with all of its foibles, should continue to scale when it moves to PC AMD hardware, because it will have been built with AMD hardware in mind from the ground up. This *should* let AMD scale well into the future where gaming benchmarks on the CPU are concerned, GPU if and when they catch up, eventually, and their continued presence on the consumer PC lets them build mindshare they can leverage back against Intel and Nvidia in the PC space. Assuming, of course, the server and mobile markets don't drop out from under AMD before their plan can come to fruition. SwissArmyDruid fucked around with this message at 06:24 on Mar 21, 2017 |
# ? Mar 21, 2017 06:20 |
|
I'm kinda surprised at the sales already on Ryzen. Seeing the R7 1700 for $285 is making it hard to wait for the R5 1600X.
|
# ? Mar 21, 2017 06:54 |
|
spasticColon posted:Which games though? There are newer games that scale very well with the number of CPU cores/threads a system has while other games only scale well with CPU clock speed regardless of the core/thread count because they are poorly optimized piles of poo poo.
|
# ? Mar 21, 2017 11:46 |
|
SourKraut posted:I'm kinda surprised at the sales already on Ryzen. Seeing the R7 1700 for $285 is making it hard to wait for the R5 1600X. Where?
|
# ? Mar 21, 2017 16:13 |
|
Lolcano Eruption posted:Where? lmgtfy https://www.reddit.com/r/buildapcsales/comments/60i4e0/cpu_amd_ryzen_7_1700_8core_processor_30_ghz_37ghz/
|
# ? Mar 21, 2017 16:21 |
|
SwissArmyDruid posted:Generally speaking: Any console games whose development began after Ryzen's launch, and also receives a PC port. This makes no sense, the CPUs in the consoles have no relation to the core layout/architecture of new Zen chips. Other that having > 4 cores.
|
# ? Mar 21, 2017 16:35 |
|
Gwaihir posted:This makes no sense, the CPUs in the consoles have no relation to the core layout/architecture of new Zen chips. Other that having > 4 cores. Why couldn't the Zen microarchitecture be used for a new APU for the next console generation?
|
# ? Mar 21, 2017 17:30 |
|
There's no reason it couldn't, but it isn't used in any console now, and no games that have been developed for the existing consoles are going to magically know about it's unique core arrangement or idiosyncrasies.
|
# ? Mar 21, 2017 17:47 |
|
Gwaihir posted:This makes no sense, the CPUs in the consoles have no relation to the core layout/architecture of new Zen chips. Other that having > 4 cores.
|
# ? Mar 21, 2017 17:52 |
|
The specifics are relatively easy to tweak ad long as a game is written with a reasonable degree of threading in the first place. Taking a game from single-threaded or quad-threaded to 8+ threads is the problem, not so much the exact core type. In fact the specifics are arguably the business of the OS in the first place. A game really shouldn't care about its underlying processor more than "number of cores" and a rough performance index per core and total. Having to lasso threads to specific cores is not optimal nor desirable. It's really the job of the scheduler not to bounce threads all over in the first place. Paul MaudDib fucked around with this message at 17:57 on Mar 21, 2017 |
# ? Mar 21, 2017 17:53 |
|
Obsurveyor posted:Why couldn't the Zen microarchitecture be used for a new APU for the next console generation? It's not being used right now, and it's not being used in upcoming Xbox "Scorpio" project which will be to the Xbox One as the PS4 Pro is to the PS4. That means at least 3 or 4 years from now until any Zen architecture is showing up in new consoles, and even more lag time from Zen-optimized console games making it to PC. There's no guarantee that the next consoles will even tap AMD for the CPU, they could switch to an Intel solution if Intel's willing to offer the right incentives.
|
# ? Mar 21, 2017 18:04 |
|
How do you know Scorpio won't use Zen? I can't find any confirmation of the CPU choice other than rumors that's it's not Jaguar. Jaguar is getting really long in the tooth at this stage and seems like a poor choice for a shiny new high-end console released in late 2017.
MaxxBot fucked around with this message at 18:32 on Mar 21, 2017 |
# ? Mar 21, 2017 18:29 |
|
MaxxBot posted:How do you know Scorpio won't use Zen? I can't find any confirmation of the CPU choice other than rumors that's it's not Jaguar. Jaguar is getting really long in the tooth at this stage and seems like a poor choice for a shiny new high-end console released in late 2017. Lisa Su has said, multiple times, that custom Zen SoC's would not be available until 2018. So unless she's been deliberating misleading investors and we get a surprise Zen in Scorpio, it seems rather unlikely.
|
# ? Mar 21, 2017 18:34 |
|
And then you take that 2018, and assume the worst case scenario possible. So, rather than actual 2018, FY2018, so, some time in early 2019.Gwaihir posted:This makes no sense, the CPUs in the consoles have no relation to the core layout/architecture of new Zen chips. Other that having > 4 cores. You're not thinking long-term enough. AMD is in at the ground floor with console games, they can hypothetically get devs to optimize this bit of the game in this way that uses work spread out across six threads, as opposed to one that's packed into four, if it would benefit both parties. Such an arrangement would clearly benefit a chip with more cores, over one with 10% better single-threaded perf. Are AMD cutthroat enough to do this? It remains to be seen, but AMD holds a pretty strong position, being able to work with developers on a large chunk of mainstream games before Nvidia can get in and go, "Psst. Hey. Use Gameworks poo poo here. None of that OpenCL crap." SwissArmyDruid fucked around with this message at 18:53 on Mar 21, 2017 |
# ? Mar 21, 2017 18:45 |
|
MaxxBot posted:How do you know Scorpio won't use Zen? Because games wouldn't be able to handle the differing layout of cores and general CPU aspects? The whole point of "Scorpio" is to be like the PS4 Pro in that all software will run correctly without any performance penalty, while most software can be run a bit faster and some software will be updated to take full advantage of the new power. Current games on the Xbox One rely on not ecountering a bottleneck for communication between the cores of the type the current Zen CPUs have between cores 0-3 and cores 4-7. Unless AMD secretly has Zen CPUs with 8 cores all on the same thing unlike the CPUs they've actually released which split 8 cores up into two seperate 4 core units, I don't see how that's possible to implement. On a normal PC OS environment, these differences wouldn't matter much - the OS and drivers would smooth over the differences. But in the console world where things are getting programmed to match very specific hardware against an OS that does much less, it can cause all sorts of havoc. Beautiful Ninja posted:Lisa Su has said, multiple times, that custom Zen SoC's would not be available until 2018. So unless she's been deliberating misleading investors and we get a surprise Zen in Scorpio, it seems rather unlikely. There's this too, but the immediate issue of software compatibility is more pressing. fishmech fucked around with this message at 20:05 on Mar 21, 2017 |
# ? Mar 21, 2017 20:03 |
|
fishmech posted:Because games wouldn't be able to handle the differing layout of cores and general CPU aspects? Somebody please explain why some of you people act like switching from Jaguar to Zen is like switching from PowerPC to X86? They're both 8 core x64 architectures. Optimizing for Vega instead of GCN is probably going to be a bigger deal.
|
# ? Mar 21, 2017 20:22 |
|
May I interest you fine folks in crashing Ryzen machines even from a VM? https://www.techpowerup.com/231536/amd-ryzen-machine-crashes-to-a-sequence-of-fma3-instructions
|
# ? Mar 21, 2017 20:27 |
|
No Gravitas posted:May I interest you fine folks in crashing Ryzen machines even from a VM? from the linked thread: TheStilt posted:The issue with Flops was found and fixed in the beginning of february.
|
# ? Mar 21, 2017 20:32 |
|
No Gravitas posted:May I interest you fine folks in crashing Ryzen machines even from a VM? I would... except that IOMMU groups are still le epic hosed.
|
# ? Mar 21, 2017 20:38 |
|
eames posted:Somebody please explain why some of you people act like switching from Jaguar to Zen is like switching from PowerPC to X86? Nobody said that, where did you even pull that out of your rear end? Just being 8 core x64 architectures is hardly good enough to ensure strict 100% compatibility in the console realm. Again, it's fine in PCs because the OS handles far more hardware abstraction, the way console do things you can't expect to go all the way from the Jaguar APUs to a theoretical and non-existent Zen-based APU and expect everything to work correctly at the same speed or better.
|
# ? Mar 21, 2017 20:39 |
|
I've seen multiple rumors that Scorpio will at least ditch the 32MB of ESRAM, if not switch to Zen or semi-Zen or something else radical so I dunno, either the rumors are stupid and it really is just going to be a speed boost with more GPU units like the PS4 Pro, or Microsoft is going to risk a few non-compatible titles or force the publishers to issue patches or something. I mean, usually you can count on "rumors are stupid".
|
# ? Mar 21, 2017 21:06 |
|
|
# ? May 30, 2024 13:14 |
|
Well, the Xbox One also is literally Windows 10, so whatever they do to Fix Zen to work on Desktop Windows 10, potentially can be backported to fix whatever changes Zen would be from Jaguar. Overall though I agree that there is going to have to be some sort of hidden APU in the works that may or may not be Zen for the Scorpio going off the talk and numbers they are saying it is going to hit. 4K and VR are going to be a big thing they are going to try and capture for it for sure.
|
# ? Mar 21, 2017 21:39 |