Amiga.org
Amiga computer related discussion => Amiga Hardware Issues and discussion => Topic started by: Piru on May 04, 2012, 01:15:57 AM
-
Lower bar means better result:
(http://sintonen.fi/pics/blender_ppclinux_benchmark.png)
sources:
AmigaONE X1000 1.8GHz: http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=35671&forum=34&start=60&viewmode=flat&order=0#664189
PowerMac G5 MP 2.3GHz: http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=35671&forum=34&start=60&viewmode=flat&order=0#664202
PowerBook G4 1.67GHz: my own test run
-
Very cool Piru!
You gave them what they asked for.
Its noteworthy that the X1000 beats a single thread G5.
I still would have liked a dual G4 in that mix.
-
it should be noted that Blender on the X1000 is running with software rendering as full hardware acceleration is currently not available under Debian PPC so the numbers in the graphs will only get better :-)
-
Its noteworthy that the X1000 beats a single thread G5.
Unless I'm misreading the chart, the X1000 dual-threaded beats a single-threaded G5; that big-ass bar on the left is for the single-threaded X1000.
-
it should be noted that Blender on the X1000 is running with software rendering as full hardware acceleration is currently not available under Debian PPC so the numbers in the graphs will only get better :-)
It should be noted that Blender does not use graphics hardware for any kind of rendering in this test, except to blit the updated graphics to the screen while the rendering progresses.
Blender does have support for OpenCL GPU rendering though, and such results should indeed be significantly better. It wouldn't really help in comparing the CPUs however.
-
O...K.. Well...
I guess the X1000 doesn't compare well using Blender.
I'd still like to see the same test performed on a dual G4.
Something tells me that the G5 would not perform significantly better.
I guess I'll stay with my current machine (and OS).
-
No big surprice on tests so far.
PA6T performs as expected (notebook caliber chip).
So far it seems that FPU/SIMD performance is not as good as PASemi estimated. IIRC, according to PASemi PA6T should perform like G5 at the same clock rate.
PA6T was technically the best available option for custom high end PPC board, it just is too expensive and the insanely complex motherboard design makes the price even worse when compared to non AOS4 HW.
Definitely the best available option as the SMP development platform for AOS4.
Other than that, summing best parts...
Very good results in:
- fileoperations (2x speed even with old drivers vs others)
- mpeg decoding (20% faster than G4)
- RAM speed (10x speed when compared to G4s)
- slightly faster L2 than on any other
So far we have not seen tests where fast RAM access is truly put in use (lack of such SW?).
I expect to see "better than ever before" results from cards on PCIe, once drivers get beyond handling x1 speed.
update/
Blender might require some rework, as also the 2.3Ghz G5 seems pathetic against the G4... and PA6T almost as it would not be using Altivec. Was the same blender binary in use in all cases?
/update
@Iggy
"You gave them what they asked for."
Hmm???
-
What's surprising to me is the superlinear performance increase with two threads. How were the tests set up and measured? Is the difference significant or within the margin of error?
-
very poor results, indeed. unless there is a software glitch in there somewhere, I'd say it becomes increasingly clear why Apple moved on to x86. It seems PA Semi's party piece was power savings, but they forgot sbout the power
-
very poor results, indeed. unless there is a software glitch in there somewhere, I'd say it becomes increasingly clear why Apple moved on to x86. It seems PA Semi's party piece was power savings, but they forgot sbout the power
They were initially doing "G5 laptop" chip. But they had roadmap to have up to 16 cores, IIRC.
@Trev
Dualcore 970 gives 1.85x the performance of single core.
PA6T gives 2.0x the performance of single core.
Might show that PA6T has a more modern MP aware design. There starts to be some use for the high bandwidth and large cache when there are more cores to use them.
Some older tests: http://www.barefeats.com/g5sum.html
http://www.barefeats.com/g5.html
That would hint that G5 should perform better than how it did in blender tests (vs G4).
(it should be up to almost 2x better in rendering??)
-
PA6T was technically the best available option for custom high end PPC board, it just is too expensive and the insanely complex motherboard design makes the price even worse when compared to non AOS4 HW.
No, it wasn't. It's performing poor. Most tests indicate that the e600 core is pretty superior (it's really telling that even a single threaded 744x at lower clock beats a dual threaded higher clocked PA6T).
If A-Eon would have used an e600 based processor (86xx - I dont want to say "I told you so", but well...) they would have a cheaper and more powerful board with a processor that still is supported. Well, now Freescale encourages not to use 86xx for new designs any more, but that line of processors is still active and supported for a while.
I never understood the decision for the PA6T. It's an überexpensive, poor performing EOL chip. A shame so much money was wasted - what else could have been achieved with these $$$?
-
They were initially doing "G5 laptop" chip. But they had roadmap to have up to 16 cores, IIRC.
What's with your fetish for extreme amounts of cores in *desktop systems*?
:confused:
-
@takemehome...
No fetish.
Just that there is no way forward without going multicore. World knows it, every Amiga flavor OS team knows it.
For most desktop needs 2Ghz single core seems enough, while there are a lot of productivity SW that benefit from more cores as well as other advancements.
btw. PA6T was planned to have 8 cores, not 16 as I said before.
@zylesea
"No, it wasn't. It's performing poor. ..."
MPC8641D system would have been cheaper, yes.
But slower (it's slower than those powerBook G4's) in almost every aspect and with less modern technology for developers to tinker with.
IIRC, for G5 macs they managed to get up to 2x speeds in rendering with single 1.6Ghz G5 when compared to 2xG4 systems (Bryce test), so there must be some optimizations ahead before we have seen the maximum out from PA6T.
-
Blender does have support for OpenCL GPU rendering though, and such results should indeed be significantly better. It
Results with OpenCL are significantly better.
My 4 year old graphics card renders twelve times faster than the four cores in my pc.
That makes this whole benchmark stupid and pathetic trolling.
In the real world, no one will use 3D software that runs only on the CPU, and wait, when the GPU is able to render the scene almost immediately.
-
Results with OpenCL are significantly better.
My 4 year old graphics card renders twelve times faster than the four cores in my pc.
That makes this whole benchmark stupid and pathetic trolling.
In the real world, no one will use 3D software that runs only on the CPU, and wait, when the GPU is able to render the scene almost immediately.
You are missing the point, so don't be a dick about it. The benchmark makes a perfetly valid comparison of the different architectures. The fact that you can do it faster on a graphics card is completely irrelevant when stacking these numbers up against eachother.
If we are going to approach this from a "real world" angle, you might as well just download a pre-rendered picture of the test scene in question, since that will probably be more effective than most other solutions.
-
@ppcamiga1
Is there OpenCL benchmark where the effect of system bandwidths are observed?
For example PCIex4 vs PCIex16.
-
@ppcamiga1
Is there OpenCL benchmark where the effect of system bandwidths are observed?
For example PCIex4 vs PCIex16.
Maybe so.
I did not do such tests.
I have in my PC slowest PCI Express 1x16 ver 1.0.
The is significant difference in rendering time on the GPU in comparison to the CPU.
I might change my motherboard with a faster but for now what I have is enough for me.
-
... nothing worth quoting ...
It's very simple.
Once again yet another who has a problem with Amiga hardware (sexy but slightly more expensive than PC), start with useless benchmarks.
"so don't be a dick about it. "
and accept this simple explanation.
-
Results with OpenCL are significantly better.
My 4 year old graphics card renders twelve times faster than the four cores in my pc.
That makes this whole benchmark stupid and pathetic trolling.
In the real world, no one will use 3D software that runs only on the CPU, and wait, when the GPU is able to render the scene almost immediately.
Only a complete *IDIOT* would confuse a CPU benchmark with GPU benchmarks, and come to the conclusion "pathetic trolling" based on this... :crazy:
Crazy!
:lol:
-
In the real world, no one will use 3D software that runs only on the CPU, and wait, when the GPU is able to render the scene almost immediately.
*raises hand* I do. Both because POV-Ray doesn't support OpenCL to begin with, and because Intel chipset video sucks and it'd probably be slower even if it was supported.
But, y'know, that doesn't support your argument, so feel free to claim that it totally doesn't count.
-
@Iggy
"You gave them what they asked for."
Hmm???
From the first thread:
With linux will be more interesting benchmark with LAme, Blender and Mplayer...
-
@Piru
Sundown said that the test was 6 seconds faster under OS4. It would be interesting to see the test result from your Powermac G4 under MorphOS. Since MorphOS is generally faster than OS4 it may well be a good bit faster than Linux for the same test.
-
@thread
Interesting observation by pavlor:
http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=35671&forum=34&start=100&viewmode=flat&order=0#664371
"Edit - exact numbers:
PowerBook G4 1667 MHz, Blender 2.49: 325 s
PowerBook G4 1667 MHz, Blender Piru: 205 s
Edit2 (Blender 2.48 - same as used on OS4):
PowerBook G4 1500 MHz, Blender 2.48a: 425 s
AmigaOne XE G4 1000 MHz, Blender 2.48: 591 s"
AOS4 users would be happy to get those Blender optimizations, as it seems to make Piru's Blender 2x faster.
-
It's very simple.
Once again yet another who has a problem with Amiga hardware (sexy but slightly more expensive than PC), start with useless benchmarks.
"so don't be a dick about it. "
and accept this simple explanation.
You didn't really explain anything. It's not a matter of having a problem with Amiga hardware as much as it is an informative benchmark. The numbers speak for themselves, and if they offend you, that's your problem.
I don't see how you can call someone a troll for posting a benchmark of hardware that could all potentially run Amiga-like operating systems just because the numbers aren't in your favor. If the truth hurts, learn to deal with it yourself instead of calling people trolls. If you think the benchmarks are useless, go ahead and don't mind them, but don't pretend that it's not an indication of CPU power.
-
*raises hand* I do. Both because POV-Ray doesn't support OpenCL to begin with, and because Intel chipset video sucks and it'd probably be slower even if it was supported.
But, y'know, that doesn't support your argument, so feel free to claim that it totally doesn't count.
Intel Ivybridge supports OpenCL.
From http://news.povray.org/povray.general/thread/%3Cweb.4e2d1fcce45434ac53ab8e5e0@news.povray.org%3E/
CPU using 8 threads
fps 33.39695
2.9942 seconds
OpenCL CPU using 8 worker units
Intel(R) Core(TM) i7 CPU 950 @ 3.07GHz
2.8005 seconds
OpenCL GPU using 18 worker units
Cypress
1.7871 seconds
----
OpenCL can slightly speed up the CPU path.
-
Only a complete *IDIOT* would confuse a CPU benchmark with GPU benchmarks, and come to the conclusion "pathetic trolling" based on this... :crazy:
Crazy!
:lol:
Why the separation between CPU and GPU when AMD GCN supports IEEE 754-2008 floating point operations (including double precision FP)?
-
Results with OpenCL are significantly better.
My 4 year old graphics card renders twelve times faster than the four cores in my pc.
That makes this whole benchmark stupid and pathetic trolling.
In the real world, no one will use 3D software that runs only on the CPU, and wait, when the GPU is able to render the scene almost immediately.
Yes, the whole thing is totally pointless. You already got the scene rendered, so what's the point of rendering it again on two other systems? You already got the picture you wanted! :rolleyes:
-
AOS4 users would be happy to get those Blender optimizations, as it seems to make Piru's Blender 2x faster.
This is the blender version from my linux distribution (debian unstable). I installed it with "apt-get install blender"
It was not built from source nor was any kind of optimizations applied.
-
Intel Ivybridge supports OpenCL.
Oh, I'm sure; I just meant that with the general performance of Intel GPUs I wouldn't be surprised to find out that CPU-only rendering was faster anyway :lol:
-
This is the blender version from my linux distribution (debian unstable). I installed it with "apt-get install blender"
So you were comparing results of two completely different Blender versions (sid has v2.63 while squeeze delivers v2.49)?
Now that should have been mentioned for sure.
Could you please update your initial graph with the proper value from this page (http://www.eofw.org/bench/) (it's test no. 3786)? As pointed out by pavlor on AW, Blender 2.49 takes 325 seconds on a Powerbook matching yours - not the 205 seconds stated in your posting.
-
So you were comparing results of two completely different Blender versions (sid has v2.63 while squeeze delivers v2.49)?
Now that should have been mentioned for sure.
Could you please update your initial graph with the proper value from this page (http://www.eofw.org/bench/) (it's test no. 3786)? As pointed out by pavlor on AW, Blender 2.49 takes 325 seconds on a Powerbook matching yours - not the 205 seconds stated in your posting.
:roflmao: Good luck with that
http://www.amiga.org/forums/showthread.php?t=61493 (http://www.amiga.org/forums/showthread.php?t=61493)
Everyone knows Piru's true agenda sadly
-
:roflmao: Good luck with that
Why don't you do me a favour and stop polluting these threads with useless drivel about the profession of Piru's mom? I'm actually interested in this stuff.
Yeah, he didn't mention that the hardinfo benchmarks were single core. Big deal. Why did I not hear you complain about him not mentioning that his Powerbook - which seemed to completely destroy the dual core X1000 in the Blender benchmarks until a few minutes ago - was only using one core aswell?
Everyone knows Piru's true agenda sadly
[...]
***X1000- I BELIEVE ***
There are a hundred virgins waiting for you somewhere. I'm pretty sure they didn't lie about that!
-
Everyone knows Piru's true agenda sadly
as this is actually the case, its time to stop mentioning it every other post..
-
Why don't you do me a favour and stop polluting these threads with useless drivel about the profession of Piru's mom? I'm actually interested in this stuff.
...
I'm not interested in deceitfulness & trolling and I hope his mom isn't either:razz:
-
as this is actually the case, its time to stop mentioning it every other post..
a simple clarification on the first posts of his 'benchmark' threads to tell the truth about them is all it takes. I don't care about which car is faster by .1 milliseconds as long as all the cars in the race are racing by the same rules.
-
You don't like the benchmarks. No one likes knowing their 3,000$ "modern spec" hardware is outperformed by a 50$ ebay buy.
Ghandi said the truth is still the truth, even if people attack you for telling it, its still the truth.
People have the right to know this stuff. Even if people have hidden agendas as some have suggested, it dosn't matter. Its still the truth.
-
A technical point also...
The processor in this x1000 was originally designed to be a mobile or laptop cpu for apple to compete with g4 speeds and abilities.
a 100$ -150$ ebay buy of just about any g5 mac blows it out of the water.
Please morphos, add g5 support. I will register that day...
-
A technical point also...
The processor in this x1000 was originally designed to be a mobile or laptop cpu for apple to compete with g4 speeds and abilities.
a 100$ -150$ ebay buy of just about any g5 mac blows it out of the water.
Please morphos, add g5 support. I will register that day...
I'd second that call as I am buying some components from AmigaDave to complete a 2.5 GHz G5 Power mac.
And as far as agendas go, you OS4 fans are the real conspirators in this jihad.
When some one presents a fact that you don't favor, you go low and get personal.
Harry hasn't done anything but present real world data.
If that bugs you, its your problem.
Thank God David still talks to me so I can get a balanced opinion from someone who owns and X1000 and a couple MorphOS systems.
-
You don't like the benchmarks....
Who doesn't like the benchmarks??
Compare apples to apples and present facts is all I'm saying.
Don't care if you're comparing PC's or cars !
Don't care about Blue vs Red unlike some people here.
-
According to shown benchmarks of this thread I wonder how anyone could see G5 as a good option for Amigalike OS. If it was true, that G4 performs better than liguid cooled G5, why would anyone want G5 over G4? For house heating purposes?
Some PowePC Results from fastest to slowest:
115s blender 2.43 G5 2.3Ghz Dualcore (erroneously reported singlecore?)
117s blender 2.44 G5 2.3Ghz Dualcore
117s blender 2.44 G5 1.3Ghz Dualcore (erroneously reported singlethread?)
120s blender 2.49 G5 2.3Ghz Dualcore
121s blender 2.49 G5 2Ghz Dualcore (erroneously reported singlecore?)
127s blender 2.49 G5 3Ghz Dualcore (erroneously reported singlecore?)
202s blender 2.5a PowerBook 1667Mhz
205s blender latest PowerBook 1667Mhz (Piru)
206s blender 2.48? x1000 1800Mhz Dualcore
223s blender 2.49 G5 2.3Ghz
234s blender 2.49 G5 2.1Ghz single core
235s blender 2.49 G5 3Ghz SingleCore
256s blender 2.42 PowerBook 1667Mhz
305s blender 2.43 PowerBook 1667Mhz
318s blender 2.44 PMac 1600Mhz
325s blender 2.49 PowerBook 1667Mhz
331s blender 2.49 PowerBook 1667Mhz
368s blender 2.43 MacMini 1420Mhz
420s blender 2.48? x1000 1800Mhz single core
425s blender 2.48a PowerBook 1500Mhz
540s blender 2.48 MacMini 1420Mhz
591s blender 2.48 A1XE 1000Mhz
1180s blender 2.48 SAM440 800Mhz (AOS4)
And when we observe with blender versions...
Then with blender 2.49 or 2.48 & dualcore:
120s blender 2.49 G5 2.3Ghz Dualcore
121s blender 2.49 G5 2Ghz Dualcore (erroneously reported singlecore?)
127s blender 2.49 G5 3Ghz Dualcore (erroneously reported singlecore?)
206s blender 2.48? x1000 1800Mhz Dualcore
Then with blender 2.49 or 2.48 & singlecore:
223s blender 2.49 G5 2.3Ghz single core
234s blender 2.49 G5 2.1Ghz single core
235s blender 2.49 G5 3Ghz SingleCore
325s blender 2.49 PowerBook 1667Mhz
331s blender 2.49 PowerBook 1667Mhz
420s blender 2.48? x1000 1800Mhz Single core
425s blender 2.48a PowerBook 1500Mhz
540s blender 2.48 MacMini 1420Mhz
591s blender 2.48 A1XE 1000Mhz
1180s blender 2.48 SAM440 800Mhz (AOS4)
With same blender version G4 is a slower than G5, but faster than PA6T.
Performance scaling etc.:
G5-2300single vs G4-1667 - 1.3 x Mhz - 1.4 x result - G5 better
G5-2300single vs PA6T-1800single - 1.28 x Mhz - 1.88 x result - G5 better
G4-1667 vs PA6T-1800single - 1.08 x Mhz - 0.77 x result - G4 better
G4-1667 vs PA6T-1800dual - 1.08 x Mhz - 1.6 x result - PA6T is better
The last part shows PA6T advantage, one day, after AOS is updated to use both cores.
-
According to shown benchmarks of this thread I wonder how anyone could see G5 as a good option for Amigalike OS. If it was true, that G4 performs better than liguid cooled G5, why would anyone want G5 over G4? For house heating purposes?
It has two cores, thus performing much better when rendering is split into two threads (as you can see from the results), just like the new Amiga computer chipset. Then again, some people say that Amiga-like operating systems will only ever run on single cores if they want to retain some level of backwards compatibility. I have no idea if this is true, though.
-
...Then again, some people say that Amiga-like operating systems will only ever run on single cores if they want to retain some level of backwards compatibility. I have no idea if this is true, though.
To my understanding every Amigaflavour has plans to go multicore in the future (with and without sandboxing, etc...).
I hope it happens during my lifetime....
(I vote for compatibility break initially (perhaps nonSMP and SMP versions of kernell available in parallel for a while) while building the sandbox during longer period of time, if the compatibility need remain)
-
@KimmoK
I'm not sure I get your point.
From your own quotes, the G5 comes off looking pretty good.
-
@KimmoK
I'm not sure I get your point.
From your own quotes, the G5 comes off looking pretty good.
Yes.
But before that it seemed like G4 is as fast as single core G5. That's what I meant and that's why I collected more results to see more clearly.
More detailed study reveals that G5 actually is not that bad. Better than what I expected.
Also, in Bryce tests that Barefeats once did it seems that with optimizations G5 can be far ahead of G4. (up to 2x faster per core, even per Mhz, IIRC).
Time will tell if that kind of optimizations can be done on 32bit SW on AOS for PA6T.
-
It has two cores, thus performing much better when rendering is split into two threads (as you can see from the results), just like the new Amiga computer chipset.
There's also the matter of memory speed; the G4s (at least in Macs) only ever had up to a 167MHz non-DDR FSB. Even the slowest G5 machines have much faster front-side buses than that, so anything memory-intensive is going to see a significant boost on the G5.
-
G5 FSB...
Was it 32in + 32out hypertransport kind of bus?
For example on 2Ghz G5 the FSB ran at 1Ghz (2x4GB/s).
While G4 can do 1.3GB/s(one direction at a time) and PA6T 32 GB/s (IIRC 16+16).
-
According to Wikipedia (http://en.wikipedia.org/wiki/PowerPC_970#Design):
The processor has two unidirectional 32-bit double data rate (DDR) buses (one for reads, the other for writes) to the system controller chip (northbridge) running at one quarter of the processor core speed. The buses also carry addresses and control signals in addition to data so only a percentage of the peak bandwidth can be realized (6.4 GB/s at 450 MHz). As the buses are unidirectional, each direction can realize only half the aggregate bandwidth, or 3.2 GB/s.
So not exactly optimal (why oh why is there still address/data multiplexing in this day and age!?) but a damn sight better than the G4.