Iirc, there's a Killzone 2 presentation that has a similar breakdown for the Cell usage, and it lists 4 SPUs being used. We know for a fact devs had access to 6 of the 7 SPUs as that's how it was in Linux, and everyone was much more open about CPU usage back then. It was brought up that maybe the extra 2 SPUs were being used for audio mixing and compression (for Dolby 5.1 and DTS output), and I'm reminded that someone mentioned audio acceleration wasn't up and running for that KZ:SF demo, so that could've taken a couple of cores too. Obviously I don't have links for any of this so I could be wrong, but I'm at work and don't have time to look it up.
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Substance Engine benchmark implies PS4 CPU is faster than Xbox One's
- Thread starter Brad Grenz
- Start date
serversurfer
Member
Looks like it was running on five of the available six, but your point still stands. Shadow Fall running on six cores tells us devs have a minimum of six cores available, not a maximum of six.
Anyone else tweet Shu about this? I did, but he never responds to me. ;.;
andromeduck
Member
well technically kinnect is still way better than the PS Eye for 3d as well as audio
outside of that, pretty much...
it's pretty hilarious really
This presentation explicitly says 1.6GHz, and it's dated November 19:
http://www.slideshare.net/DevCentralAMD/mm-4085-laurentbetbeder
http://www.slideshare.net/DevCentralAMD/mm-4085-laurentbetbeder
Thanks for digging that up. It's strange since I recall there not being any mention of audio, but there it is on the sixth SPU along with various system tasks. Are audio or system tasks listed on the KZ:SF pdf?
Well, AMD and someone from SCEA sound like solid sources.
I'm leaning towards it being an extra core available for now.Cool. Have not seen this before. Definitely sounds like an extra core then. Doesn't explain this benchmark though.
Does anyone know if optimizing for 7 cores would be difficult since it's not an even split?
didnt the ps3 have 7 cores? Maybe its a superstitious trend with them,
THE SEVENTH CORE OF THE SEVENTH GEN!!!
Well now the 8th gen but they have the 8gb of ram for that.
Sword Of Doom
Member
Yes it's the same CPU but we also have OS overhead to consider. So it's very simplistic to just say they're the same and call it a day
serversurfer
Member
Weird. This seems to be a per-core test, as discussed, so what gives? If PS4 is 1.6 GHz and XBone is 1.75 GHz, that would mean that PS4 is 28% faster per-clock. How could that possibly be?
Insane Metal
Member
I guess no one remembered that the PS4 CPU also has some modifications on it. There's an interview where Cerny talks about it. There are some 'pluses' to it.
Many factors, the Xbone API might be not very optimized, so this test program that they ran compiled less optimized and ran slower for example.
Or the 3 OS virtualization overhead actually prevents using a core to 100% on the Xbone, for example.
Dear god, if these are the actual real reasons then imagine if no upclock did happen...
"Weak laptop CPU"
Hahaa!!
serversurfer
Member
It seems unlikely those modifications would really be relevant here. They're for improving GPGPU performance, but this is strictly a CPU test; Substance can already run directly on the GPU if desired, so I doubt it's hybrid code or anything like that.
That'd be pretty unoptimized
So, it reserves two full cores plus 22% from the other six? That seems even less likely. I'd be more inclined to believe 22% from all eight cores, but that seems like kind of a shitty way to do things; I'd think devs would rather just have six cores to themselves rather than having the OS trodding on every single thing they try to do, ever.
NDAs suck. I hate not knowing stuff.
That'd be pretty unoptimized
So, it reserves two full cores plus 22% from the other six? That seems even less likely. I'd be more inclined to believe 22% from all eight cores, but that seems like kind of a shitty way to do things; I'd think devs would rather just have six cores to themselves rather than having the OS trodding on every single thing they try to do, ever.
NDAs suck. I hate not knowing stuff.
I don't know man, I'm just guessing like everyone else who's not a dev. Your guesses are just as valid as mine.
Doesn't sound like the kind of software that would have to talk to APIs a lot, and I wonder what x86 compiler Sony could use that's that much faster than MS's. Does Intel still make their heavy duty optimizing compiler? I get the funny feeling the critical sections probably use some assembly on x86 anyway.
inpHilltr8r
Member
Google "PS4 LLVM" if that was a typo. Or just read slide 7 of Laurent's presentation.
serversurfer
Member
Oh, I know. Sorry, I didn't mean to come off like a dick or otherwise overly critical of your response.
It's more that I was assuming a basic level of competency on both sides of the fence, so I was assuming any major performance difference would be spec related. I mean, your suggestions basically amount to, "Perhaps MS are just really, really bad at making consoles." I'm certainly no fan of MS or their products, but even I have a hard time imagining this is all simply down to their incompetence. =/
KoolAidPitcher
Member
That was only speculation on my part.
That is indeed interesting news to me. Glad to see the PS4 compiler has C++x11 support. The std::shared_ptr should allow developers to develop using the RAII paradigm more easily rather than having to sandwich (for every malloc you must free, for every new you must delete). Problems with sandwiching, for example, an exception being thrown and resources not being properly freed are the source of most memory leaks in my experience.
Another of my favorite C++x11 features is std::thread and thread_local storage. The clock speed of processors is not increasing at the rate it used to; however, processors are getting more parallel, so these are obviously awesome standardized features for parallelizing your code. In order to make my code thread-safe, I use std::lock_guard and std::unique_lock; however, they are not sometimes not the most efficient lock available and in a tight loop, I might opt for something more efficient. Yes, I am a huge fan of RAII.
Obviously, most of these features were part of the boost C++ library prior to being standardized; however, I like having lots of helpful tools in my standard library. Now I only wish the standard library had tribool, and a couple of other boost features.
flying dutchman
Neo Member
Is Sony the first console manufacturer to have not released the clock speed of their CPU?
Nintendo didn't say anything about their Wii U clocks either. They were revealed when some hackers did some of their magic.
these systems have been out for a month now. Why havent hackers picked them apart yet ...
backbreaker65
Banned
It's as if you didn't read the disclaimer.
12. DISCLAIMER & ATTRIBUTION The information presented in this document is for informational purposes only and may contain technical inaccuracies, omissions and typographical errors. The information contained herein is subject to change and may be rendered inaccurate for many reasons, including but not limited to product and roadmap changes, component and motherboard version changes, new model and/or product releases, product differences between differing manufacturers, software changes, BIOS flashes, firmware upgrades, or the like. AMD assumes no obligation to update or otherwise correct or revise this information.
There's more at the link.
Muppet of a Man
Member
Done and done. Nothing else to see here, folks.
andromeduck
Member
1. even that doesn't make much sense as both the PS4 and XB1 contain 2 CPU modules in their APU and this data doesn't really make sense in that scenario... if they were just talking of the platform as a whole then they wouldn't have said 1 CPU
the most likely scenario is then that the PS4 is clocked at 2 GHz
as for 6 vs 7 cores it's really not as bad once you get beyond the first few, dev's have gotten a lot better at it in recent years
It makes plenty of sense. The most likely scenario is the PS4 being clocked at 1.6GHz because thats what the Sony/AMD source says.
serversurfer
Member
So? That's just legal drivel acknowledging the possibility they made some errors in the slides, but there's no reason to think they did.
serversurfer
Member
apart from the fact the PS4's CPU outperforms XBone's per-core, of course.
Which could be for a number of reasons unrelated to clock speed.
serversurfer
Member
Maybe Allegorithmic stuffed a bunch of peanut butter in to the XBone before they ran the test.
Does anyone have the PDF? I want to get a closer look at the Processor on page 3 I wonder if that's the PS4 design or just something they used for the presentation?
it look a lot like this one
Edit: never mind
andromeduck
Member
14/12 = 1.16667
1.6 * 7/8 = 1.4
1.75 * 6/8 = 1.3125
1.4/1.3125 = 1.06667
it still doesn't add up if you assume 1.6 GHz on the PS4 unless you assume all 8 cores are available as
(1.6*8/8)/(1.75 * 6/8) = 1.2190
given the evidence, we can reasonably conclude the PS4's CPU clock is well above 1.6GHz which leaves us with two possible scenarios:
1. by 1 CPU they meant 1 CPU core (as in CPU_0, CPU_1 etc.) which is the most reasonable, clocked at 2.0
1.75 * 14/12 = 2.04
2. by 1 CPU they meant 1 APU AND OS/apps occupy 1 core, CPU clocked at 1.75 just like the Xbone's
1.75*7/8/(1.75 * 6/8) = 1.1666
it really couldn't
both APU's use the exact same CPU modules - same cache size same interface same everything
the only thing that is different is again memory but in this case that shouldn't be an issue as a workload like this should be fairly easy on the cache and if anything DDR3 should perform better due to the marginally lower latency and absence of a bandwidth bottleneck
KoolAidPitcher
Member
Maybe whatever the hell this is boost the CPU up by 25% when running the Substance Engine because it's some kinda co-processor or accelerator?
I have no idea what that chip does, but if I had to speculate based only on it's location on the board, I tend to believe it might be a hardware-based DRM solution to make PS4 software more difficult to emulate on a virtual machine.
Because the PS4 uses an X86_64 instruction set, Sony might have thought it would be a whole lot easier than previous console generations to create a virtual machine capable of running PS4 software. Want to know what the most efficient type of emulation is? Emulation that you don't have to do, because your processor already supports the instruction set.
I have no evidence of this; however, it is technically possible that the instructions for PS4 software both on the HDD and on blu-ray could be scrambled/encrypted, and that chip is a hardware-based solution to on-the-fly decrypt the stream. We already know the PS4 does on-the-fly decompression using a zlib decoder, so this would not necessarily be a crazy idea.
Once again, I really don't know anything about that chip. This is only my hypothesis.
GoofsterStud
Member
How accurate is GopherD?
zomgbbqftw
Banned
So this is the chatter I heard recently on Sony's 8GB bombshell.
In early 2012 SCE were told by developers that their 4GB target was too low and that they needed a minimum of 8GB for next gen development to be competitive over a 5-7 year period. In the middle of 2012 Sony approached Samsung over the viability of low power 4Gbit chips and were willing to give seed money for investment to that end. The rest as they say is history...
It just surprises me that Microsoft engineers didn't look at this path after revealing to developers that their system would have a pitifully small amount of bandwidth.
Seems pretty clear ms themselves bought into the cloud thing. Its really the only thing that makes sense considering they just flat ignored technology, then tried to tell all of us how computers work.
My uneducated guess would probably involve considering the overall budget of the console as well, mainly due to the "Kinect in every box" idea, and maintained costs as a priority rather than developers' wishes or concerns.
Panajev2001a
GAF's Pleasant Genius
Does anyone have the PDF? I want to get a closer look at the Processor on page 3 I wonder if that's the PS4 design or just something they used for the presentation?
it look a lot like this one
Edit: never mind
Those slides are very interesting and kind of confirm that the secondary ARM processor and the DPU we see mentioned are quite the same thing basically. Documentation in those slides and from Tensilica offer quite a lot of nice tidbits.
Reading those slides made me think about sound processing on the GPU and it's presented WHY's and WHY NOT's... I seem to recall people talking about AMD's audio technology built in PS4's GPU, yet this SCEA paper does not seem to make any mention of that when discussing the pros and cons of running game audio code on CPU vs GPU vs ACP/Secondary Processor... at an AMD conference nonetheless.
Curious don't you think?
Globalisateur
Banned
If PS4 has 7.5 cores available for the games then those 12/14 benchmark results are roughly compatible with:
- 7.5 cores at 1.6ghz on PS4
- 6 cores at 1.75ghz on X1
It's the most probable cause of those results, in my opinion, because we would not need any overclocking on the PS4 CPU, the benchmark would still be right (with Matt post) and this UbiSoft dev tweet would also be accurate:
Because the whole CPU of the X1 is still stronger than PS4 CPU but you can get more of the CPU on PS4 for the games.
- 7.5 cores at 1.6ghz on PS4
- 6 cores at 1.75ghz on X1
It's the most probable cause of those results, in my opinion, because we would not need any overclocking on the PS4 CPU, the benchmark would still be right (with Matt post) and this UbiSoft dev tweet would also be accurate:
Because the whole CPU of the X1 is still stronger than PS4 CPU but you can get more of the CPU on PS4 for the games.
Well, I would not describe Xbone bandwith as not pitifully small. That can be onlu be used for WiiU now.
[~15GB/s for its ram]68GB/s DDR3 was fine for their vision of the system, and was again fine for a battle against competitor with 4GB of GDDR5 ram [out of whitch 0.5-1GB would go to OS].
That Cerny bomb must have been painful for them.
This makes sense if the benchmark is multi threaded.If PS4 has 7.5 cores available for the games then those 12/14 benchmark results are roughly compatible with:
- 7.5 cores at 1.6ghz on PS4
- 6 cores at 1.75ghz on X1
It's the most probable cause of those results, in my opinion, because we would not need any overclocking on the PS4 CPU, the benchmark would still be right (with Matt post) and this UbiSoft dev tweet would also be accurate:
Because the whole CPU of the X1 is still stronger than PS4 CPU but you can get more of the CPU on PS4 for the games.
serversurfer
Member
Except based on the Tegra and A6 results, the test seems to be running on a single core, making all of the mental gymnastics about half-core OS reservations and the like irrelevant.
Or, maybe Tegra is super bad at this algorithm.
Or, maybe A6 is super good at it.
Or, maybe someone needs to break some NDAs.
Interesting:
(Worth mentioning though that you cannot judge the overall performance of a CPU from its performance on "intrinsics".)
Also mentions "hUMA" several times.
Tensilica the maker of the DPU are the one's who make the audio / video chips that are part of the AMD APU's & GPU's.
Their DSP's are also a part of the Xbox One.
It is a weak laptop cpu, hence Sony loaded up on compute too.
Unfortunately game devs seem to have a bad case of Not Implemented Here, and tend not to trust the standard libs and advanced C++ features.
If you dig up the Sony LLVM presentation they say that their compiler defaults to no exceptions and no RTTI.
But I think hope that that'll start to change this generation.
ThirdMartini
Neo Member
There is another possible explanation for the performance difference. MS have many times mentioned they run 3? different OS on the xbox one at once in a virtualizer. (IE: likely some form of their Hyper-V virtualizer)
Well if the test was run inside a VM container on top of Hyper-V it will not get the same performance as being run natively. Hyper-V performance is anywhere between ~70% and ~95% of native. ( Depending on the type of compute ) We do not know how the PS4 OS runs code either, but its possible that code run on PS4 does NOT run in a full virtualizer but instead some sort of sandboxing, which would yield native hardware performance.
So even if running on a faster cpu the benchmark could run slower in a VM. With my tinfoil hat on... if I realized that I'm taking a performance hit because I'm forcing games to run in a virtualizer... I might up the CPU clock rate to insure performance parity with my competitor that is not.
If this is the case, it certainly is possible and even likely that REAL/TUNED for a VM code would ACTUALLY end up faster on the Xbox One in the future.
Well if the test was run inside a VM container on top of Hyper-V it will not get the same performance as being run natively. Hyper-V performance is anywhere between ~70% and ~95% of native. ( Depending on the type of compute ) We do not know how the PS4 OS runs code either, but its possible that code run on PS4 does NOT run in a full virtualizer but instead some sort of sandboxing, which would yield native hardware performance.
So even if running on a faster cpu the benchmark could run slower in a VM. With my tinfoil hat on... if I realized that I'm taking a performance hit because I'm forcing games to run in a virtualizer... I might up the CPU clock rate to insure performance parity with my competitor that is not.
If this is the case, it certainly is possible and even likely that REAL/TUNED for a VM code would ACTUALLY end up faster on the Xbox One in the future.
Codeblew
Member
LOL. I wonder if PS4 uses GCC or Clang. I am sure they had to modify the compiler whichever they chose.
Similar threads
- Poll