I'm curious to see how your VII performs over time. I also got a VII and was able to run a dozen or so work units on it, but ran into the same problem that I had with my Vega56: It seems to get through some units just fine but then runs into others that cause display driver crashes/restarts or outright system hangs. The same system runs work units on a R9-280x and simultaneously mines on the Vega (now VII) for weeks on end, so I don't think power is the issue. Anyway, I'm planning to try crunching under Linux this weekend.

I posted earlier in the thread on exactly that comparison. A Radeon VII versus a Titan V comparison over at MilkyWay. The Radeon VII still won.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I posted earlier in the thread on exactly that comparison. A Radeon VII versus a Titan V comparison over at MilkyWay. The Radeon VII still won.

Application differences.
For double precision the Vega VII is rated at 3.5 TFLOPS, the Titan V at up to 6.875 TFLOPS. Both are short of their workstation lineage hardware.
In the Single precision comparisons I've seen, the Vega VII is behind even GTX 1080 TI Founders Edition), although from the run times so far here on Seti it looks to be pretty close to the GTX 1080Ti.
As I said, application differences.

Here Nvidia rules due to the Special Application, were there to be similar development for AMD hardware then they would have similar performance (within architecture limitations- just as certain games have always favour certain hardware architectures, GPU compute work is no different).

---

Grant
Darwin NT

Ok i decided to be the guinea pig for this after all.
The first few work units seem to run twice as fast as on my old GTX 1080ti.
I need to run more workunits to be sure.

+1

---

A proud member of the OFA (Old Farts Association).

I'm curious to see how your VII performs over time. I also got a VII and was able to run a dozen or so work units on it, but ran into the same problem that I had with my Vega56: It seems to get through some units just fine but then runs into others that cause display driver crashes/restarts or outright system hangs. The same system runs work units on a R9-280x and simultaneously mines on the Vega (now VII) for weeks on end, so I don't think power is the issue. Anyway, I'm planning to try crunching under Linux this weekend.

Would it be possible to run both cards on the same system at the same time? For Seti?

---

A proud member of the OFA (Old Farts Association).

Faster than a RTX 2070 on the SoG app.
https://setiathome.berkeley.edu/workunit.php?wuid=3356955178

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Faster than a RTX 2070 on the SoG app.
https://setiathome.berkeley.edu/workunit.php?wuid=3356955178

About 10 seconds faster than my GTX 1070s, the difference between mine & the one you quoted is the one you quoted is running stock with no command line. The Radeon & My GTX 1070s are making use of command line options.
The Radeon would be faster still with period_iterations_num= set to 1.
So I guess that puts it around the GTX 1080/ GTX 1080Ti performance levels for Seti.

---

Grant
Darwin NT

Up and crunching now under Ubuntu 18.04. Vega 56 and Radeon VII in the same system.

https://setiathome.berkeley.edu/show_host_detail.php?hostid=6993114

I applied your command line parameters. I am now completing WUs on Radeon VII in a little over 3 minutes.

Are you running the opensource driver or the AMD driver?

I applied your command line parameters. I am now completing WUs on Radeon VII in a little over 3 minutes.

I think you can "merge" the two entries with Windows and your faster gpu into a single userid.

Tom

---

A proud member of the OFA (Old Farts Association).

The only issue I am seeing is that the driver crashes when i run more than 1 WU on the same card. Despite trying a number of parameters, I have been unable to prevent this from happening. I think I could get much higher throughput with multiple WUs. Wondering if you have any ideas. I have tried adjusting voltages and clocks on the card to no effect. Similarly, changing command line switches has not been effective.

Right now, I am not merging the two profiles because I dont want the scheduler to decide to send me the wrong type of work units.

When I put together my pc last month I started with an rx580. Ended up returning due to poor Linux drivers in Ubuntu. I got myself 2 Gtx 1070ti’s and they each seem to be happy knocking out units every 2.5-3.5 minutes. Amd Linux support is terrible. I built another Linux box with Ubuntu, the inbuilt Vega has no Linux drivers.

Linux support for AMD is actually very good. But you have to know what drivers to install. They certainly don't make it as easy as the Windows approach of point and click on an official driver page.

You have to choose between the AMDGPU drivers and the ROCm drivers depending on the kernel and distribution you are using.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I tried the drivers of the amd website. Tried amdgpu and the amdprogpu, both crashed x windows. As for the Ryzen 3 with the Vega, there are no drivers. Amd community thread basically points to the open Mesa drivers. I ended up putting the gt710 in that system, that way my son could play his games that’s require shader 3.0 support. Nvidia made it so easy to install drivers.

I know that the Radeon VII is usable for crunching with the available drivers since it is the card preeminent at Einstein@home. Maybe go over there and read through the many Radeon Vega threads for clues about which drivers are working.
https://einsteinathome.org/content/all-things-radeon-vii-vega-20

https://einsteinathome.org/content/radeon-vega

https://einsteinathome.org/content/quick-guide-how-install-opencl-amd-gpus-linux-kubuntu-1804-and-similar-distro?page=1

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Was just wondering how the radeon vii performs. My 2080ti does 183kflop tasks, 3 tasks at 8min20sec. Or roughly 2:45 per 183kflop task. How does the R VII do??

From looking at hosts with the Radeon VII, it looks very similar. I found one Windows host that runs tasks in around the same time and one Linux host that runs tasks a little faster.
https://setiathome.berkeley.edu/show_host_detail.php?hostid=7933012
https://setiathome.berkeley.edu/show_host_detail.php?hostid=6993114

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Hi All,

I bought a Radeon VII for use in an external GPU enclosure (eGPU) connected to my 2019 macbook pro via thunderbolt 3.

The card works in that I can get a desktop to appear on a connected monitor (if I dont connect a monitor the card doesnt even show up).

However, when running BOINC, Seti at Home, it doesnt add the extra GPU. My macbook has two GPUs built in (Intel 630 and Radeon Pro Vega 20). Those work fine in BOINC. But, the 3rd GPU (Radeon VII) doesn't show up at all.

Any ideas?

Thank you,

Mark

---

Your times are already very impressive.
If you want to speed up a little bit look for a file called mb_cmdline_win_x86_SSE2_OpenCL_ATi_HD5.txt.
Its located in your projects folder.

Add the following line

-sbs 2048 -period_iterations_num 10 -high_perf -spike_fft_thresh 4096 -tune 1 64 1 4 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 64 -oclfft_tune_cw 64 -cpu_lock

Save as text.
On next WU start it will use the optimized values.

Mike

Can i add this line to all mb_cmdline...ati...txt files?
I have 4 of them.

I also own a VII now and after amazing performance jumps at Einstein@Home during the Pentathlon now i try to use it for Seti@Home. But the stock performance is in my eyes very poor with Ubuntu.
2 WUs parallel get a daily output of around 30.000cr.
3 WUs parallel get 40.000cr.
My 1070ti reaches 60.000cr with one WU per CUDA100.

I hope that the commandline-parameter will increase this a little bit.

---