All things Radeon VII / Vega 20

Chooka
Chooka
Joined: 11 Feb 13
Posts: 134
Credit: 3689865759
RAC: 2207346

Well here we are nearly 1

Well here we are nearly 1 week on. For me, the undervolting has worked well (touch wood)

My errors have dropped off with no new ones and my run times have remained the same. This has been much more stable than reducing the power limit. My GPU fan sits at 1400 -  1440 rpm now. I can't even hear it.

Interesting numbers there Gavin. Your times are about 357 seconds (about 178/sec/wu) running X2.

Mine are 528 seconds (176/sec/wu) running X3

Your CPU time is much lower than mine (93 sec vs my 118 seconds) I'm not sure how much difference that makes.

I also shutdown crunching for a few hours each weekend while I squeeze in some gaming :D I should try running X2 again just to see how it goes.


Gavin
Gavin
Joined: 21 Sep 10
Posts: 191
Credit: 40644337738
RAC: 6

Chooka wrote: Interesting

Chooka wrote:

Interesting numbers there Gavin. Your times are about 357 seconds (about 178/sec/wu) running X2.

Mine are 528 seconds (176/sec/wu) running X3

Your CPU time is much lower than mine (93 sec vs my 118 seconds) I'm not sure how much difference that makes.

Technically your machine is quicker but as I said in my earlier post (and shown be your numbers) the gain running x3 is so so marginal it is not worth the effort and impact on the wallet! I still wholeheartedly recommend you switch to running the card x2.

Be careful when comparing your runtimes to mine, remember my seven year old quad core platform is trying to tame two Radeon vii's and as such there is a performance hit to be factored into my results... Not sure how to quantify that though.

Is your machine running more than this project? That could impact throughput as much as your gaming needs. Switch to running x2, crunch only at Einstein, buy yourself an Xbox and claim that top single card spot in the top hosts list!!!! - I'm kidding of course but it would be nice if it was so simple :-)

Gav.

 

Chooka
Chooka
Joined: 11 Feb 13
Posts: 134
Credit: 3689865759
RAC: 2207346

I had that issue again the

I had that issue again the other day where I downloaded and updated to the latest Radeon drivers, only to find that when I started BOINC again, it showed Device 0 & Device 1 again.

The only way I fixed it was to do a system restore. I'm really not sure why it does that!

Thank goodness for a system restore but gee I'd like to update to the latest drivers! Anyone else have this issue with a Radeon VII driver update? (Win 10)


Chooka
Chooka
Joined: 11 Feb 13
Posts: 134
Credit: 3689865759
RAC: 2207346

Hi Gavin, I will try X2

Hi Gavin,

I will try X2 again eventually. Life's busy atm. Currently via GPU-Z my average power draw is 192.8W.

I really need to change my Vega 56's! I think they are currently drawing more than this Vega VII. And yes, my machine has most cores crunching Universe@Home. I've balanced reduced cores vs output. As in, I could reserve more CPU cores and get a slight GPU bump but I've sacrificed the GPU bump so I can crunch more WU's. I think I have 3 or 4 threads free to crunch my X3 Einstein WU's.

Thanks for your advice :)

ps...no Xbox here....I'm too uncoordinated at those controllers ;)


Stefan Ledwina
Stefan Ledwina
Joined: 23 Oct 05
Posts: 17
Credit: 2492232233
RAC: 1422969

Chooka schrieb:I had that

Chooka wrote:

I had that issue again the other day where I downloaded and updated to the latest Radeon drivers, only to find that when I started BOINC again, it showed Device 0 & Device 1 again.

The only way I fixed it was to do a system restore. I'm really not sure why it does that!

Thank goodness for a system restore but gee I'd like to update to the latest drivers! Anyone else have this issue with a Radeon VII driver update? (Win 10)

 

I always  have that problem when I update the drivers.

To fix that I, use DDU (https://www.wagnardsoft.com/) to completely uninstall the old driver, and then install the new one...

Chooka
Chooka
Joined: 11 Feb 13
Posts: 134
Credit: 3689865759
RAC: 2207346

Yeah last time I tried that,

Yeah last time I tried that, it was a nightmare. I basically had to wipe C:drive and start again. Even my system restores failed.

Thx anyway.


Chooka
Chooka
Joined: 11 Feb 13
Posts: 134
Credit: 3689865759
RAC: 2207346

Ah Ha! I haven't had any

Ah Ha! I haven't had any errors lately but my runtimes have increase quite a bit over the week. I'm not sure if that's just a change in WU's or because I've changed CPU projects.

I just went to reboot my PC (because that fixes everything as we know) and I saw that error I was referring to. Not sure if it's linked to the lack of output from the VII every so often.

 Posting pics isn't easy here, so I'll describe it.

Title of the error is - RadeonSettings: RadeonSettings.exe - Application Error.

"The instruction at 0x00007FF9AF3D9680 reference memory at 0x0000000200000007. The memory could not read."

 

I've seen this most times when I restart my PC after the card hangs. Might mean more to one of you than me.


Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 4957
Credit: 18619847825
RAC: 5614109

Radeon VII has been

archae86
archae86
Joined: 6 Dec 05
Posts: 3157
Credit: 7205854931
RAC: 928988

As of a few minutes ago,

As of a few minutes ago, Nowinstock only shows available Radeon VII stock at Newegg, which asserts it has available inventory of the Asrock model (and not at an inflated price).

It will be very interesting to see how well the Navi models perform on Einstein relative to the fabulous performance here of the Radeon VII, which was out of all proportion to the relative gaming performance of that model.

As the Navi models are materially different architecturally and use a very different type of memory, I think only actual Einstein trials will tell us much of how well or how poorly they perform here on the current Gamma Ray Pulsar application--let alone on future applications.

 

cecht
cecht
Joined: 7 Mar 18
Posts: 1525
Credit: 2867826018
RAC: 2058557

archae86 wrote:As the Navi

archae86 wrote:
As the Navi models are materially different architecturally and use a very different type of memory, I think only actual Einstein trials will tell us much of how well or how poorly they perform here on the current Gamma Ray Pulsar application--let alone on future applications.

I'm looking forward to someone starting an "All Things Navi 10" forum topic. :)

Ideas are not fixed, nor should they be; we live in model-dependent reality.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.