MP 1,1-5,1 cMP UserBenchMark X5675 RX580

[andysa1966]

https://www.userbenchmark.com/UserRun/18152154

cMP all up to date

I ran the userbenchmark running Windows 10 pro, CPU score was very good

Performing way above expectations (88th percentile)

GPU score good, however the GPU is stated as

Performing way below expectations (12th percentile)

Is this normal in cMP?

 

[orph]

It looks like you have a dual CPU cmp with an RX580 and a SSD, is something wrong with it?
or just showing of benchmarks.

 

[andysa1966]

It looks like you have a dual CPU cmp with an RX580 and a SSD, is something wrong with it?
or just showing of benchmarks.

RX580 performing way below expectaions?

My question is regarding the low 12th percentile performance of the RX580 in my cMP, is it normal in all cMP?

 

[h9826790]

Make sure Windows set to run at high performance, but not best power saving.

Make sure the GPU’s power draw is not limited in the AMD driver control panel,

Run the benchmark three times, still consistently at 12 percentile?

 

[orph]

it's not a benchmark i have seen before so i don't know if it is one you can trust.

for osx the valley benchmark is a good option
https://benchmark.unigine.com/valley

for windows id use the superposition benchmark
https://benchmark.unigine.com/superposition

ill check my card now to give you something to compare to

ps what RX 580 do you have?

osx valley benchmark with a RX 580 pulse 1330mhz, when you do it make sure to use the extreme HD preset and not to change any settings just hit the benchmark button



superposition benchmark windows 10 RX 580 pulse 1300mhz (make sure you have the same settings as me)

 

[Pentaxer]

How have you powered your GPU?

Do you use the latest drivers?

Have you installed all windows system updates?

 

[andysa1966]

it's not a benchmark i have seen before so i don't know if it is one you can trust.

for osx the valley benchmark is a good option
https://benchmark.unigine.com/valley

for windows id use the superposition benchmark
https://benchmark.unigine.com/superposition

ill check my card now to give you something to compare to

ps what RX 580 do you have?

osx valley benchmark with a RX 580 pulse 1330mhz, when you do it make sure to use the extreme HD preset and not to change any settings just hit the benchmark button
View attachment 849710
View attachment 849711

superposition benchmark windows 10 RX 580 pulse 1300mhz (make sure you have the same settings as me)
View attachment 849713
View attachment 849714

Its a Saphire Pulse 4GB, Ill give that bench mark a go

 

[andysa1966]

Make sure Windows set to run at high performance, but not best power saving.

Make sure the GPU’s power draw is not limited in the AMD driver control panel,

Run the benchmark three times, still consistently at 12 percentile?

Ill try agin over the weekend
[doublepost=1564028805][/doublepost]

How have you powered your GPU?

Do you use the latest drivers?

Have you installed all windows system updates?

GPU is powered using with two into one cable (2x6 into 8 i think) that came with the cMP when i bought it. Yes its all up to date, a fresh install with latest updates for windows 10 and AMD drivers

 

[Pentaxer]

The bench GPU results are affected not only by GPU, but also by CPU, motherboard, memory and so on.
cMP is an 11 years old creature:
1) very old CPU (~ 1 generation of Core technology, now the 9th is in use);
2) it doesn't support PCI-E 3.0;
3) it has low memory bandwith and high latensy in some cases due to CMP (dual CPU).

Most results in benchmark are related to newer systems.

Don't look at bench, just use it to make money or fun. cMP does it's work very well

 

[orph]

yes id asume the GPU is working just fine, id gess the benchmark you used is just funky

 

[andysa1966]

The bench GPU results are affected not only by GPU, but also by CPU, motherboard, memory and so on.
cMP is an 11 years old creature:
1) very old CPU (~ 1 generation of Core technology, now the 9th is in use);
2) it doesn't support PCI-E 3.0;
3) it has low memory bandwith and high latensy in some cases due to CMP (dual CPU).

Most results in benchmark are related to newer systems.

Don't look at bench, just use it to make money or fun. cMP does it's work very well

Sure it not for serious work, i was just interested if that was the case as you describe or something needed tweeking, Ilike tinkering

 

[andysa1966]

it's not a benchmark i have seen before so i don't know if it is one you can trust.

for osx the valley benchmark is a good option
https://benchmark.unigine.com/valley

for windows id use the superposition benchmark
https://benchmark.unigine.com/superposition

ill check my card now to give you something to compare to

ps what RX 580 do you have?

osx valley benchmark with a RX 580 pulse 1330mhz, when you do it make sure to use the extreme HD preset and not to change any settings just hit the benchmark button
View attachment 849710
View attachment 849711

superposition benchmark windows 10 RX 580 pulse 1300mhz (make sure you have the same settings as me)
View attachment 849713
View attachment 849714

Seems about right

 

[andysa1966]

Make sure Windows set to run at high performance, but not best power saving.

Make sure the GPU’s power draw is not limited in the AMD driver control panel,

Run the benchmark three times, still consistently at 12 percentile?[/Q

Make sure Windows set to run at high performance, but not best power saving.

Make sure the GPU’s power draw is not limited in the AMD driver control panel,

Run the benchmark three times, still consistently at 12 percentile?

not much difference in the results, I did notice in the userbench mark, so I checked in AMD watt man and the GPU clock is 1150mhz, I looked up online and all examples were 1257mhz?

 

[h9826790]

Is that a used mining card? Those clock speeds do not look like native at all.

 

[andysa1966]

Is that a used mining card? Those clock speeds do not look like native at all.

perhaps, I purchased it recently from ebay

 

[orph]

if you have not changed any of the settings it looks like a used mining card the voltages are all super low & vram is OC'ed as well as low clock speed.

so at the mo it's set to be right in the efficiency curve for power to speed for compute use & as a big plus it will use less power so the fans will be near silent non stop and the vram speed will be maxed out.

if you want you can set it to default you can flash back the original rom or if you want to keep it like this, if you do want to flash it to stock make a backup of this rom just in case you like it more.

In windows i keep mine set to 1000mhz 99% of the time as it’s all i need and keeps the fans whisper silent at about 20% slower i guess

is this your card

im guess thanks to the bios mod it's name is mixed up in the benchmarks so want to double check

edit
im tempted to say the score you got in the benchmark looks good so it may be worth keeping like this & at least do backup the bios if you do want to try the orginal settings.

 

[zoltm]

not much difference in the results, I did notice in the userbench mark, so I checked in AMD watt man and the GPU clock is 1150mhz, I looked up online and all examples were 1257mhz?

I have recently acquired a piece of used Sapphire Pulse RX 570 4G for sane price, with box and warranty, moving up from the trusty Gigabyte RX 460 4G.

To be honest with you, with my casual Internet surfing workflow, I am not feeling a whole bunch of bump up in user experience. But hey, for just 50$, why not some cheap thrill.

Thanks to @h9826790 excellent article on Bootcamp, it has been my first time ever running Windows on my Apple computer since ditching windows for more than 10 years now. Playing with AMD Wattman @ windows is a lot of fun .

Thanks @andysa1966 for your sharing above on Wattman setting of your card, I tried them on my RX 570. Kind of liking it, the setting keeps the card running at low temp and low wattage. I have to give up a few points on benchmark but honestly in actual use case, you can't really feel it. I don't really game on Windows so ability to keep the card running cool and quiet is even more important for me.

Cheers. E

 

[h9826790]

I have recently acquired a piece of used Sapphire Pulse RX 570 4G for sane price, with box and warranty, moving up from the trusty Gigabyte RX 460 4G.

To be honest with you, with my casual Internet surfing workflow, I am not feeling a whole bunch of bump up in user experience. But hey, for just 50$, why not some cheap thrill.

Thanks to @h9826790 excellent article on Bootcamp, it has been my first time ever running Windows on my Apple computer since ditching windows for more than 10 years now. Playing with AMD Wattman @ windows is a lot of fun .

Thanks for your sharing above on Wattman setting of your card, I tried them on my RX 570. Kind of liking it, the setting keeps the card running at low temp and low wattage. I have to give up a few points on benchmark but honestly in actual use case, you can't really feel it. I don't really game on Windows so ability to keep the card running cool and quiet is even more important for me.

Cheers. E

For info, if you want to keep that setting in macOS. You have to flash the graphic card.

https://forums.macrumors.com/threads/sapphire-pulse-rx580-8gb-vbios-study.2133607/
[doublepost=1564386537][/doublepost]

perhaps, I purchased it recently from ebay

Then it is the problem.

The clock speed is reduced quite a bit, which of course can reduce the performance.

 

[zoltm]

For info, if you want to keep that setting in macOS. You have to flash the graphic card.

https://forums.macrumors.com/threads/sapphire-pulse-rx580-8gb-vbios-study.2133607/

@h9826790 Thanks for your note, I've indeed re-read the thread multiple times to get a better feel on the change required. I've already download the 1.7.2 version of Polaris Editor and 2.87 version of ATI Flash. I am still getting my head around how to get the relevant value into the Polaris Editor and spin a version of the BIOS for my RX 570. I am indeed quite happy with @andysa1966 setting in Windows 10. That setting get better benchmark with Furmark and MSI Kombustor. While for Geekbench, the OpenGL benchmark is worse than the default setting, however, I am fine with having the card running in a more energy efficient manner. I did also compare the WX3100 setting you suggested. @andysa1966 setting downvolt more aggressively than the WX3100 one but it does perform more highly, very interesting. I tried to crank ALL things up initially as most beginner overclocker, I am indeed enlightened when the more aggressively downvolt setting yields more performance while more energy efficient than the crank things all up setting.

 

[h9826790]

@h9826790 Thanks for your note, I've indeed re-read the thread multiple times to get a better feel on the change required. I've already download the 1.7.2 version of Polaris Editor and 2.87 version of ATI Flash. I am still getting my head around how to get the relevant value into the Polaris Editor and spin a version of the BIOS for my RX 570. I am indeed quite happy with @andysa1966 setting in Windows 10. That setting get better benchmark with Furmark and MSI Kombustor. While for Geekbench, the OpenGL benchmark is worse than the default setting, however, I am fine with having the card running in a more energy efficient manner. I did also compare the WX3100 setting you suggested. @andysa1966 setting downvolt more aggressively than the WX3100 one but it does perform more highly, very interesting. I tried to crank ALL things up initially as most beginner overclocker, I am indeed enlightened when the more aggressively downvolt setting yields more performance while more energy efficient than the crank things all up setting.

That's not necessary a more effective downvolt.

As I said in the RX580 ROM study thread. There are two loops in the PULSE ROM to pick up invalid parameters. You can set 0.85V in Wattman, but may be the actual measured voltage still 0.98V.

You should run GPU-Z to check the measured voltage when you playing around with Wattman.

 

[zoltm]

That's not necessary a more effective downvolt.

As I said in the RX580 ROM study thread. There are two loops in the PULSE ROM to pick up invalid parameters. You can set 0.85V in Wattman, but may be the actual measured voltage still 0.98V.

You should run GPU-Z to check the measured voltage when you playing around with Wattman.
View attachment 850364

Thanks for the enlightenment. It's indeed the case, here is my GPU-Z capture.

Given this information, how would you suggest myself to key in at the Polaris Editor to get a BIOS with this exact setting? It would be cool if I can try this in MacOS.

I do aware that the current Wattman setting in Windows is NOT portable to macOS as the card would still remains factory setting when booting into macOS.

I check my cards ASIC quality,

I suppose it is quite decent, so the card seems to run fine even under 1v.

I saw that you can crank your memory to 2150mhz, may I know at what voltage to achieve that? 2050 seems to be the best I can get without any artifact. My current feeling is that running the card at lower core frequency while having the best memory speed seems to achieve better benchmarks results. Is that a proper understanding?

 

[orph]

@zoltm may be best to move the talk to this topic
https://forums.macrumors.com/threads/sapphire-pulse-rx580-8gb-vbios-study.2133607/

it's geting off topic

'casual Internet surfing workflow' wont see massive gains from RX 460 -> RX 570

 

[andysa1966]

if you have not changed any of the settings it looks like a used mining card the voltages are all super low & vram is OC'ed as well as low clock speed.

so at the mo it's set to be right in the efficiency curve for power to speed for compute use & as a big plus it will use less power so the fans will be near silent non stop and the vram speed will be maxed out.

if you want you can set it to default you can flash back the original rom or if you want to keep it like this, if you do want to flash it to stock make a backup of this rom just in case you like it more.

In windows i keep mine set to 1000mhz 99% of the time as it’s all i need and keeps the fans whisper silent at about 20% slower i guess

is this your card
View attachment 850304
im guess thanks to the bios mod it's name is mixed up in the benchmarks so want to double check

edit
im tempted to say the score you got in the benchmark looks good so it may be worth keeping like this & at least do backup the bios if you do want to try the orginal settings.

Yes that looks like the same one, it is very quiet and had a low power draw in this configuration
[doublepost=1564476867][/doublepost]

For info, if you want to keep that setting in macOS. You have to flash the graphic card.

https://forums.macrumors.com/threads/sapphire-pulse-rx580-8gb-vbios-study.2133607/
[doublepost=1564386537][/doublepost]

Then it is the problem.

The clock speed is reduced quite a bit, which of course can reduce the performance.

Thanks, I have been reading your other thread, where would I find the original rom to flash back to the card if I wanted to, AMD?

 

[h9826790]

where would I find the original rom to flash back to the card if I wanted to, AMD?

TechPowerUp