@tlosm

http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=42116&forum=32#803223

_________________
PowerMac G5 2,7Ghz - 2GB - Radeon 9650 - MorphOS 3.14
AmigaONE X1000, 2GB, Sapphire Radeon HD 7700
FPGA Replay + DB 68060 at 85Mhz

@K-L

I thought his main concern was with the P5020 model?

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@BigD

No was concern that the killer app use only one core of the x5000/20 on two and one core of X5000/40 on four

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@BigD

LOL. Well you have to do more than that. On any Workbench a must have is ClickToFront. People use Amigas without it. I don't know how they survive. And of course there's the usual use Workbench to load up DirOpus and avoid Workbench from then on. Apart from that it is nice to run 68K apps directly with a double click and see it use the resources from the native OS4 OS. Then there is screen dragging and cross screen icon dragging. A night on the town there. I've tried the "cheaper" AmigaOS clones but the look and feel just isn't there. It's not the same. So expensive yes but it also feels like the real thing.

@outrun1978

I prefer to run 68k "natively" as possible and then use RunInUAE (or any other UAE) as a fallback. I would like to see a UAE build that redirects OS calls to native OS4 calls. So would be like OS4 68k emulation but sandboxed in a way. Fuller Amiga emulation but making use of OS4 resources and screens natively.

@tlosm

Quote:

tlosm wrote: @BigD No was concern that the killer app use only one core of the x5000/20 on two and one core of X5000/40 on four

Yeah, obviously with no SMP yet 3 cores are wasted on the P5040 and 1 core on the P5020 but I though with your Linux experience you said the Ram speed on the P5020 was ridiculously slow and the P5040 was quite a bit better and nearly at G5 Quad speeds?

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

@BigD

ram speed is little better on p5040 compared the p5020 but ... bad compared the quad g5

Just for remember the quad have ddr2 (4200j) 533mhz rams, x5000 ddr3 1600 rams

Stream Quad G5 : no altivec used

Quote:

STREAM version $Revision: 5.10 $ ------------------------------------------------------------- This system uses 8 bytes per array element. ------------------------------------------------------------- Array size = 10000000 (elements), Offset = 0 (elements) Memory per array = 76.3 MiB (= 0.1 GiB). Total memory required = 228.9 MiB (= 0.2 GiB). Each kernel will be executed 10 times. The *best* time for each kernel (excluding the first iteration) will be used to compute the reported bandwidth. ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 34862 microseconds. (= 34862 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Best Rate MB/s Avg time Min time Max time Copy: 2926.7 0.054736 0.054670 0.054815 Scale: 2903.6 0.055172 0.055104 0.055203 Add: 3359.6 0.071593 0.071437 0.071846 Triad: 3377.3 0.071144 0.071063 0.071237 ------------------------------------------------------------- Solution Validates: avg error less than 1.000000e-13 on all three arrays

Stream P5040

Quote:

STREAM version $Revision: 5.10 $ ------------------------------------------------------------- This system uses 8 bytes per array element. ------------------------------------------------------------- Array size = 10000000 (elements), Offset = 0 (elements) Memory per array = 76.3 MiB (= 0.1 GiB). Total memory required = 228.9 MiB (= 0.2 GiB). Each kernel will be executed 10 times. The *best* time for each kernel (excluding the first iteration) will be used to compute the reported bandwidth. ------------------------------------------------------------- Your clock granularity/precision appears to be 1 microseconds. Each test below will take on the order of 125127 microseconds. (= 125127 clock ticks) Increase the size of the arrays if this shows that you are not getting at least 20 clock ticks per test. ------------------------------------------------------------- WARNING -- The above is only a rough guideline. For best results, please be sure you know the precision of your system timer. ------------------------------------------------------------- Function Best Rate MB/s Avg time Min time Max time Copy: 1328.9 0.120552 0.120398 0.120717 Scale: 1233.6 0.130414 0.129701 0.134902 Add: 1615.4 0.149091 0.148566 0.149944 Triad: 1612.6 0.149026 0.148826 0.149314

Last edited by tlosm on 23-Sep-2017 at 09:36 AM.
Last edited by tlosm on 23-Sep-2017 at 01:20 AM.
Last edited by tlosm on 23-Sep-2017 at 01:09 AM.
Last edited by tlosm on 23-Sep-2017 at 01:08 AM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

G5 Quad dnetc
Quote:

dnetc v2.9112-521-CFR-16021317 for Linux (Linux 4.10.1). Please provide the *entire* version descriptor when submitting bug reports. The distributed.net bug report pages are at http://bugs.distributed.net/ [Sep 14 11:28:23 UTC] Automatic processor type detection found a PowerPC 970MP (G5) processor. [Sep 14 11:28:23 UTC] OGR-NG: using core #0 (KOGE 3.1 Scalar). [Sep 14 11:28:42 UTC] OGR-NG: Benchmark for core #0 (KOGE 3.1 Scalar) 0.00:00:16.17 [22,497,333 nodes/sec] [Sep 14 11:28:42 UTC] OGR-NG: using core #1 (KOGE 3.1 Hybrid). [Sep 14 11:29:01 UTC] OGR-NG: Benchmark for core #1 (KOGE 3.1 Hybrid) 0.00:00:16.17 [42,424,383 nodes/sec] [Sep 14 11:29:01 UTC] OGR-NG benchmark summary : Default core : #1 (KOGE 3.1 Hybrid) 42,424,383 nodes/sec Fastest core : #1 (KOGE 3.1 Hybrid) 42,424,383 nodes/sec [Sep 14 11:29:01 UTC] RC5-72: using core #0 (MH 2-pipe). [Sep 14 11:29:20 UTC] RC5-72: Benchmark for core #0 (MH 2-pipe) 0.00:00:16.34 [5,648,705 keys/sec] [Sep 14 11:29:20 UTC] RC5-72: using core #1 (KKS 2-pipe). [Sep 14 11:29:39 UTC] RC5-72: Benchmark for core #1 (KKS 2-pipe) 0.00:00:17.09 [6,132,322 keys/sec] [Sep 14 11:29:39 UTC] RC5-72: using core #2 (KKS 604e). [Sep 14 11:29:59 UTC] RC5-72: Benchmark for core #2 (KKS 604e) 0.00:00:17.07 [5,593,755 keys/sec] [Sep 14 11:29:59 UTC] RC5-72: using core #3 (KKS 7400). [Sep 14 11:30:19 UTC] RC5-72: Benchmark for core #3 (KKS 7400) 0.00:00:17.07 [15,767,442 keys/sec] [Sep 14 11:30:19 UTC] RC5-72: using core #4 (KKS 7450). [Sep 14 11:30:39 UTC] RC5-72: Benchmark for core #4 (KKS 7450) 0.00:00:17.06 [18,846,672 keys/sec] [Sep 14 11:30:39 UTC] RC5-72: using core #5 (MH 1-pipe). [Sep 14 11:30:58 UTC] RC5-72: Benchmark for core #5 (MH 1-pipe) 0.00:00:17.06 [4,759,204 keys/sec] [Sep 14 11:30:58 UTC] RC5-72: using core #6 (MH 1-pipe 604e). [Sep 14 11:31:17 UTC] RC5-72: Benchmark for core #6 (MH 1-pipe 604e) 0.00:00:16.10 [4,743,695 keys/sec] [Sep 14 11:31:17 UTC] RC5-72 benchmark summary : Default core : #4 (KKS 7450) 18,846,672 keys/sec Fastest core : #4 (KKS 7450) 18,846,672 keys/sec [Sep 14 11:31:17 UTC] Compare and share your rates in the speeds database at http://www.distributed.net/speed/ (benchmark rates are for a single processor core)

Dnetc P5040

Quote:

./dnetc -bench distributed.net client for Linux Copyright 1997-2016, distributed.net RC5-72 PowerPC assembly by Malcolm Howell and Didier Levet Enhancements for 604e CPUs by Roberto Ragusa RC5-72 Altivec and OGR assembly by Didier Levet Please visit http://www.distributed.net/ for up-to-date contest information. Start the client with '-help' for a list of valid command line options. dnetc v2.9112-521-CFR-16021317 for Linux (Linux 4.11.3-X5000-Tlosm-LB). Please provide the *entire* version descriptor when submitting bug reports. The distributed.net bug report pages are at http://bugs.distributed.net/ [Sep 15 15:12:52 UTC] Automatic processor type detection found a PowerPC e5500 processor. [Sep 15 15:12:52 UTC] OGR-NG: using core #0 (KOGE 3.1 Scalar). [Sep 15 15:13:11 UTC] OGR-NG: Benchmark for core #0 (KOGE 3.1 Scalar) 0.00:00:16.27 [22,245,376 nodes/sec] [Sep 15 15:13:11 UTC] OGR-NG benchmark summary : Default core : #0 (KOGE 3.1 Scalar) 22,245,376 nodes/sec Fastest core : #0 (KOGE 3.1 Scalar) 22,245,376 nodes/sec [Sep 15 15:13:11 UTC] RC5-72: using core #0 (MH 2-pipe). [Sep 15 15:13:29 UTC] RC5-72: Benchmark for core #0 (MH 2-pipe) 0.00:00:16.13 [5,940,470 keys/sec] [Sep 15 15:13:29 UTC] RC5-72: using core #1 (KKS 2-pipe). [Sep 15 15:13:48 UTC] RC5-72: Benchmark for core #1 (KKS 2-pipe) 0.00:00:16.07 [6,054,956 keys/sec] [Sep 15 15:13:48 UTC] RC5-72: using core #2 (KKS 604e). [Sep 15 15:14:08 UTC] RC5-72: Benchmark for core #2 (KKS 604e) 0.00:00:17.04 [6,182,771 keys/sec] [Sep 15 15:14:08 UTC] RC5-72: using core #5 (MH 1-pipe). [Sep 15 15:14:26 UTC] RC5-72: Benchmark for core #5 (MH 1-pipe) 0.00:00:16.48 [5,814,945 keys/sec] [Sep 15 15:14:26 UTC] RC5-72: using core #6 (MH 1-pipe 604e). [Sep 15 15:14:45 UTC] RC5-72: Benchmark for core #6 (MH 1-pipe 604e) 0.00:00:16.05 [5,709,840 keys/sec] [Sep 15 15:14:45 UTC] RC5-72 benchmark summary : Default core : #-1 (undefined) 0 keys/sec Fastest core : #2 (KKS 604e) 6,182,771 keys/sec [Sep 15 15:14:45 UTC] Compare and share your rates in the speeds database at http://www.distributed.net/speed/ (benchmark rates are for a single processor core)

Dnetc P5020



Quote:

dnetc -bench distributed.net client for Linux Copyright 1997-2009, distributed.net RC5-72 PowerPC assembly by Malcolm Howell and Didier Levet Enhancements for 604e CPUs by Roberto Ragusa RC5-72 Altivec and OGR assembly by Didier Levet Please visit http://www.distributed.net/ for up-to-date contest information. Start the client with '-help' for a list of valid command line options. dnetc v2.9107-515-CFR-09102403 for Linux (Linux 4.13.2-X5000-jm). Please provide the *entire* version descriptor when submitting bug reports. The distributed.net bug report pages are at http://bugs.distributed.net/ [Sep 15 16:56:57 UTC] Automatic processor type detection did not recognize the processor (tag: "e5500") [Sep 15 16:56:57 UTC] OGR-NG: using core #0 (KOGE 3.1 Scalar). [Sep 15 16:57:16 UTC] OGR-NG: Benchmark for core #0 (KOGE 3.1 Scalar) 0.00:00:16.36 [20,183,307 nodes/sec] [Sep 15 16:57:16 UTC] OGR-NG benchmark summary : Default core : #-1 (undefined) Fastest core : #0 (KOGE 3.1 Scalar) [Sep 15 16:57:16 UTC] RC5-72: using core #0 (MH 2-pipe). [Sep 15 16:57:35 UTC] RC5-72: Benchmark for core #0 (MH 2-pipe) 0.00:00:16.39 [5,379,355 keys/sec] [Sep 15 16:57:35 UTC] RC5-72: using core #1 (KKS 2-pipe). [Sep 15 16:57:54 UTC] RC5-72: Benchmark for core #1 (KKS 2-pipe) 0.00:00:16.48 [5,486,483 keys/sec] [Sep 15 16:57:54 UTC] RC5-72: using core #2 (KKS 604e). [Sep 15 16:58:13 UTC] RC5-72: Benchmark for core #2 (KKS 604e) 0.00:00:16.12 [5,586,407 keys/sec] [Sep 15 16:58:13 UTC] RC5-72: using core #5 (MH 1-pipe). [Sep 15 16:58:32 UTC] RC5-72: Benchmark for core #5 (MH 1-pipe) 0.00:00:16.73 [5,269,122 keys/sec] [Sep 15 16:58:32 UTC] RC5-72: using core #6 (MH 1-pipe 604e). [Sep 15 16:58:51 UTC] RC5-72: Benchmark for core #6 (MH 1-pipe 604e) 0.00:00:16.17 [5,171,342 keys/sec] [Sep 15 16:58:51 UTC] RC5-72 benchmark summary : Default core : #-1 (undefined) Fastest core : #2 (KKS 604e)

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@BigD

As you can see on the cpu performances Post

And on ram performances post

The P5040 have good integer value but where it is really bad is on ram transfer value and this is what im say from very beginning the ram value is to much bad i was expecting better performances compared the G5 but was not like this. this slow ram made system on linux choppy but dont gave issue on AmigaOS who is much more light OS compared linux

About fpu on blender benchmark
P5020 need 1 m 19s to render 2T
P5040 need 42s to render 4T
G5 Quad need 23s to render 4T

Last edited by tlosm on 23-Sep-2017 at 09:35 AM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@tlosm

Thanks for benchmarks!

FPU speed is not that bad (cca 70 % of 970MP performance per MHz), but integer speed is clearly the strong side of e5500.


I still suspect slow GFX card drivers as primary suspect of "choppy" Linux experience.

@pavlor

Quote:

I still suspect slow GFX card drivers as primary suspect of "choppy" Linux experience.

yes it was what i reported to a-eon in my one year experience "cpu perform well" the bad is on ram performances plus pcie performing really bad "you can check on hans website too x5000 goes bad compared the sam460 with same gpu.

note on linux x5000 is setup with the kernel parameter radeon.pcie_gen2=1 on g5 i cant because gen 1.


note2: now on quad im using my x5000 ssd with ubuntu mate .. and look like another level

Last edited by tlosm on 23-Sep-2017 at 10:42 AM.
Last edited by tlosm on 23-Sep-2017 at 10:42 AM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@all



the unboxing

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@tlosm

"Like usual there is not my name.. **** A-EON"

A little childish don't you think ?

Evert

_________________
-- Posting for charity -- Investing €10 in a charity related to education or civil rights for every message I post --

*** can be what you like

Love
Hell
F**k
S**t

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome

@tlosm

Quote:

tlosm wrote: *** can be what you like Love Hell F**k S**t

Oh right ...

Evert

_________________
-- Posting for charity -- Investing €10 in a charity related to education or civil rights for every message I post --

@tlosm

Quote:

you can check on hans website too x5000 goes bad compared the sam460 with same gpu

Both of those systems provide four PCI-E lanes to their video card slots, the SAM 460 is Gen1.1 PCI-E while the X5000 is Gen2 PCI-E.
SO, the X5000 has TWICE the bandwidth of the SAM460 devoted to video.

I totally respect Han de Ruiter, BUT the X5000 has the potential to complete dust the SAM460 or Tabor in respect to video card performance.

And, from what I've seen from Freescale's documentation, the P5020 and P5040 have a MUCH higher potential memory bandwidth than the current benchmarks for the X5000 are producing.
This would lead me to conclude that there are firmware issues that can be improved upon.

In any case, except for possibly the X1000, the X5000 should outperform ALL previous OS4 platforms from Eyetech AmigaOnes, to the Pegasos2, as well as the SAM440 and SAM460.

AND, AmigaOS (or for that matter MorphOS) is NOT available for PCI-E G5 PowerMac.
Further, the X5000 was not designed primarily as a Linux platform.
Also, while I use my quad G5 for Linux, my Ryzen based X64 PC would make a MUCH better platform for Linux.

IF we are to continue to post benchmarks of unsupported hardware, would you care for me to post Ryzen 1800X benchmarks for comparison to your G5 benchmarks?
I can assure you, the disparity would be as high (and in many cases HIGHER) than what you are showing with your X5000/G5 comparisons.

To conclude, we AREN'T a Linux forum. I'm sorry you didn't get the support you think you were due from Aeon or AmigaKit. BUT I'd be willing to bet that they'd offer you a full refund to STFU and go away.

This is an Amiga forum, the X5000 AmigaOne IS an Amiga computer, and some of us would like to see the SOLE company producing AmigaOne systems survive and prosper.
And, btw, I think Matt Leaman is a bit of a knob myself (not as bad as some players in the market, but support is not his forte).
But I have known Paul Gentle longer than I've known Trevor Dickinson, and I hold both of them in the highest regard.
AND, we in the US are lucky enough to have Aaron Smith at Amiga On The Lake who (unlike Matt) is GREAT at supporting his customers (both NG AND legacy).

So, if you're not with us, you're against us.
You bias before you changed your mind was as virulent as your current negative bias.
And I'm not sure I trust either position you've adopted.

Sure, the quad G5 has a few advantages. But I own one, and I'm STILL looking forward to owning an X5000/40.

Last edited by Beans on 23-Sep-2017 at 06:15 PM.

_________________
Amiga! "Our appeal has become more selective"

http://hdrlab.org.nz/benchmark/gfxbench2d/Reference/

MemCopy

This is a series of tests to measure the speed at which data can be copied between main memory (RAM) and graphics card RAM (VRAM). A lot of graphics are not generated by the graphics card itself, but are images or videos loaded from disk or downloaded from the internet. With videos in particular, the transfer speed of data to the graphics card is an important factor.

The following memory copy tests are performed:

Copy to VRAM - A block of data is copied from RAM to VRAM. In this test, this task is performed by the CPU with no DMA

WritePixelArray - A block of data is copied from main memory (RAM) to VRAM using the Operating System's (OS') graphics library. The graphics may or may not use DMA in order to accelerate the copy operation

Copy from VRAM - A block of data is copied from VRAM to RAM. In this test, this task is performed by the CPU with no DMA

ReadPixelArray - A block of data is copied from VRAM to RAM using the Operating System's (OS') graphics library. The graphics may or may not use DMA in order to accelerate the copy operation

The MemCopy score is the weighted sum of the tests listed above. The CPU based tests are both multiplied by 0.25 while WritePixelArray and ReadPixelArray are multiplied by 0.75. More importance is placed on using the graphics library's copy functions.
------------------------------------------------------------------------------------------------------------------------------
Han's comment in this thread .

http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=42094&forum=33&start=60&viewmode=flat&order=0#802491

Last edited by Spectre660 on 23-Sep-2017 at 07:01 PM.
Last edited by Spectre660 on 23-Sep-2017 at 06:34 PM.

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

X1000 (DMA)
http://hdrlab.org.nz/benchmark/gfxbench2d/OS/AmigaOS/GraphicsCard/567/Result/2028

MemCopy Score: 1,783.79 (Hide Raw Data)
Operation MiB/s
Copy to VRAM 446.69
Write Pixel Array 1,329.17
Copy from VRAM 28.37
Read Pixel Array 890.86

Sam460ex (DMA)
http://hdrlab.org.nz/benchmark/gfxbench2d/OS/AmigaOS/GraphicsCard/522/Result/1937

MemCopy Score: 602.57 (Hide Raw Data)
Operation MiB/s
Copy to VRAM 305.42
Write Pixel Array 629.77
Copy from VRAM 53.19
Read Pixel Array 54.13

X5000/20 (No DMA)
http://hdrlab.org.nz/benchmark/gfxbench2d/OS/AmigaOS/GraphicsCard/522/Result/1989

MemCopy Score: 584.76 (Hide Raw Data)
Operation MiB/s
Copy to VRAM 533.04
Write Pixel Array 532.17
Copy from VRAM 53.71
Read Pixel Array 51.93

Last edited by Spectre660 on 24-Sep-2017 at 05:06 AM.

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@Spectre660

Why is the test for the X5000 done without DMA?

The copy to vram is quite a bit higher than the SAM460, but the other figures are only on par, or slightly slower.

As the bandwidth of the video slot is higher, the figures of the X5000 should be higher than this, which would leave me to conclude that DMA is a significant issue.

When you combine this higher bandwidth with the much better integer performance of the P5020, there should be a bigger performance disparity.

BTW - The X1000 figures are petty outstanding. Copying to and from vram is a bit slow, but those figures for MemCopy and Write/read pixel array? Wow!

16 PCI-E lanes, even if they are only Gen1 really seem to make a big difference. This should give the X1000 a two fold improvement over the X5000, and a four fold improvement over the SAM460 in overall bandwidth.

_________________
Amiga! "Our appeal has become more selective"

@Beans

i dint compared a G5 or X5000 with a Ryzen ... is like compare a fiat 500 (g5) piaggio Ape (X5000) with the last lamborghini (Ryzen)

Last edited by tlosm on 23-Sep-2017 at 07:04 PM.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome