Ethereum & OpenCL: ROCm vs. AMDGPU-PRO 17.40

[phoronix]

Phoronix: Ethereum & OpenCL: ROCm vs. AMDGPU-PRO 17.40

Following this week's Ethereum and OpenCL benchmarks with Radeon vs. NVIDIA using the latest Linux drivers, some premium supporters requested a fresh AMDGPU-PRO vs. ROCm comparison. So here are a couple of those OpenCL benchmarks of AMDGPU-PRO vs. ROCm on different Polaris / Fiji and Vega GPUs.

http://www.phoronix.com/vr.php?view=25425

[gsedej]

I am getting 21MH/s on rx480 on rocm (vs 16 on benchmark) same as michaels amgdgpu
No special bios or overclock.
Memory is at default 2000mhz cire is lowered to 900mhz due to noise and no perf inpact

[Marc Driftmeyer]

WARNING: Just a frustration observation and patience wearing thin

Serious question? How effing long will it take for ROCm that was ``supposed to be upstreamed'' to actually get upstreamed? It's not a legal obstacle anymore. This is a simple decision by AMD to appease enterprise clients first. Any half wit can figure out that if AMD has half the brain mass to proclaim to be experts at OpenCL they'd show it in any reasonable level that is respected and approaches that of Apple and macOS which has OpenCL integrated throughout the OS and all of their professional and consumer apps. They added Blender support for the sake of show casing it to sell Ryzen and GPUs only for sanctioned Linux systems. BFD.

You want to be FOSS and sell hardware, get the damn software ready or keep losing sales to that POS green company. You want to play across the consumer space, then stop begging for the Enterprise and bending over backwards for them.




Until that highlighted section says something beside HPC and UltraScale Compute all I see is, ``never going to be upstreamed,'' with OpenCL and Mesa.

Spare me the bs lines about legal wrangling. POCL has 2.x support and it's a poor man's operation, yet it doesn't show up for Blender because Blender has hardcoded OCL just for AMD and it's proprietary stack.

Wake me when the contracts run out for Ubuntu 16.04 and we all blink to see it changed to Ubuntu 17.04 and still not OpenCL 1.2/2.x making Vega/Navi and the rest a waste of effing time.

I don't see this crap with Windows which I absolutely effing loathe. I know I won't with OS X as any AMD GPU will have OpenCL because you folks won't be the one writing the code.

I don't give a rusty f*** about gaming improvements. This is LINUX. It'll pee in the wind 20 years from now and still have 1% Steam user support. Get the OpenCL open already. It's just utterly embarrasing.

You have very few FreeSync capable Displays yet on the market. Yet, OpenCL 1.2 a 5+ year old spec is still absent?

As a person w/ nearly 3K shares of AMD stock keep beating your estimates and my long bet makes me money. As a professional needing OpenCL for application use, not custom development needs, and not being a slave to Ubuntu/CentOS or some other sanctioned crap I am looking forward to the iMac Pro and just letting my Linux box go to waste while its GPUs are knee capped by poor management decisions.

The more time you waste getting every little feature parity with Windows but ignore the OpenCL end you lose more people interested in using Linux for independent production work and see those investments turn elsewhere.

In short, I'll pay $5k for thew new iMac Pro and put down another $2k in Pro Apps and more because I know the stuff works and will be supported and have a high life cycle before obsolescence. I cannot say the same for Linux or Windows workstations/desktops/laptops, especially when the software stacks are hampered.

No one but a few paying Phoronix loyalists must give a rusty f*** about Mining. Tell them to go run their own tests. People have screamed for 18 months about OpenCL and nothing new seems to be coming. The kernel is borked from AMD with ROCm and hasn't been fixed. It's got issues up the wazzu on their Github branch w/ nearly 0 feedback. This leads one proverbally back to blowing up their current Linux distribution and all your efforts to rebuild a system with a `sanctioned' distribution. Pass.

https://community.amd.com/community/devgurus/

Nothing but complaints about the software stack that is proprietary on Ubuntu and other Linux sanctioned distributions. It's nothing but spotty support for Professional only cards. That crap needs to shore up or just admit you have no plans of becoming an end-to-end solutions provider, but just a chip option.

It's like the same 5 people assigned to the forums hog the board about obscure crap that has no impact on anyone but custom OpenCL applications for enterprise clients.

Meanwhile, Mesa is insanely packed with devs getting everything under the sun leaving half the power of AMD GPGPUs sitting to rot because we've gotta get those old OpenGL games more optimizations, right? All 1% or less of the Steam community, right?

[bridgman]

If it helps, there are probably quite a few more devs working on OpenCL than are working on Mesa. The Mesa devs just happen to be extraordinarily good, which skews perception but not reality.

AFAIK most of the KFD code has been aligned with upstream and pushed out for review; we just didn't make the 4.14rc6 cutoff for 4.15.

There is also work going on to increase the OpenCL support level.

[r1348]

Please stop this madness. I want to be able to buy a gaming video card at a reasonable price.

[Kano]

Marc Driftmeyer

Why don't you use Nvidia cards for opencl/cuda? I do not understand that you seem to need use AMD hardware just because you invested some money. This is a completly independent incident. If your solution is to buy a Mac to get a better opencl stack that's fine but somehow it seems that you have got weird ideas to solve issues...

[jstefanop]

Marc Driftmeyer why do you blame amd for upsteaming problems THEY CAN NOT CONTROL... they just can not control what Linus Torvalds is doing... in my point of view they do the best they can... really.

I think most of his frustrations are from the wasted OpenGL efforts over the years on linux, with OpenCL efforts only recently gaining steam. I agree with him, no one seriously games on linux, and with all the effort wasted on getting 3D on par with windows over the years, they could have built an amazing linux compute software stack that cold be rivaled by no one (you could even say ROCm is almost at the point even though is only a year or so old).

The lack of any control over our GPUs is the most frustrating part on the compute side. We are still no where near fglrx's level of control (full CLI controls over clocks/voltages). They have exposed basic clocking over /sys but this is usually buggy, and if you want real control have to hack the open source bits to do what you want which is supper annoying.

I want to be an expert in optimizing the compute applications I run, not an expert in figuring out how AMD's driver works...thats AMD's job, and thats why we pay 1K+ for GPUs. I literally had an AMD dev just point me to the linux kernel and pretty much say here its open source now so code what you want it to do. If this is this stance AMD takes with their open source initiative (i.e. offloading driver work they should do to end users)...it wont be a good outcome...

[r1348]

why do you not just dual use your hardware? gaming sure and if you do not do gaming just do some mining instead.
This is profitable even on relative high electric bill costs. for example my 2 systems with 6 vega-64 make 200€ electric costs per month but they make a 310€ output per month.

Take a look at this very nice list: http://www.worldatlas.com/articles/e...the-world.html
I'm right there at #1.
Also, I'm not really interest in fairy money that can crash at any given moment, or be easily manipulated by a few "whales". Not to mention its environmental cost.

[lucrus]

but they work hard and plan this for Linux kernel 4.15....

Did that happen? I haven't kept myself up to date with Rocm news, and I'm courious to know if current kernels have those bits mainlined.

[bridgman]

Did that happen? I haven't kept myself up to date with Rocm news, and I'm courious to know if current kernels have those bits mainlined.

IIRC support for most GPUs went into 4.17, Vega10 support went into 4.18, and Raven Ridge support is lined up for 4.19.