Posts by Shaggie76

1) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 2018929)
Posted 14 Nov 2019 by Profile Shaggie76
Post:
Thanks for the update. I have been scouring the used market for a 1080, and am blown away that that 1660ti goes nearly toe to toe with it. Very impressive. Rethinking the path forward, that's for sure.
And although it's not on the list yet (since it's just been released) the GTX 1660 Super is meant to be on par (or even slightly better) than the 1660Ti, and it's cheaper.

I just checked and in the last scan there were four GTX 1660 SUPERs -- I'll give it a few weeks and scan again.
2) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 2018703)
Posted 12 Nov 2019 by Profile Shaggie76
Post:
The RTX 2060 and 2070 SUPERs are included now -- some excellent performance for both of them depending on what you're looking for.

Also some Radeon Pro RX 500X series which don't seem particularly exciting.

This graph is getting too long -- maybe I should start culling the R7, R9, GTX 400 and GTX 500 cards?

3) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 2017858)
Posted 5 Nov 2019 by Profile Shaggie76
Post:
I ran a scan over night and there aren't enough SUPER cards in circulation to qualify yet so the results are more or less the same as before. I'll do another scan in a week or two.
4) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1997206)
Posted 7 Jun 2019 by Profile Shaggie76
Post:
I noticed that the GTX 1660 (non-Ti), GTX 1650, and even the Radeon VII are not on this chart. I'm assuming there is not enough of a sample size to include them, or is there another reason?

Also, if/when the GTX 1650 is put on the chart, how will you compute the Credit/Watt-Hour for this card since some run power off the motherboard and others have a power connector?

Yeah, not enough 1650's yet. There might of been enough Radeon VIIs but I couldn't find TDP specs for them so I left that for the next scan since I was out of time.

Since I can't measure actual power draw I cross-reference the published TDP specs from Wikipedia assuming that all vendors and cards will have approximately the same variation in overclocked parts etc.
5) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1997116)
Posted 7 Jun 2019 by Profile Shaggie76
Post:
New this scan, Radeon RX 590, GeForce GTX 1660 Ti, and GeForce RTX 2060: the 1660 Ti looks to be a new leader in performance/watt and the 2060 looks like an excellent economy option!

6) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1988200)
Posted 31 Mar 2019 by Profile Shaggie76
Post:
I just wish those easy to calculate numbers would be plotted with red X on the chart too.

As a programmer I'm sure that you're familiar with Amdahl's law; making the app 3x faster for a dozen users is insignificant compared with making the app faster for the 90,000 other active users volunteering for the project. Please stop asking me to show how much faster your code is on my graphs until it's an official app that everyone can use.
7) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1988037)
Posted 30 Mar 2019 by Profile Shaggie76
Post:
New scan includes early results for the GTX 1660 Ti -- only 14 hosts so there's more variance but it's looking very good.

8) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1984636)
Posted 12 Mar 2019 by Profile Shaggie76
Post:
New this scan: NVIDIA GeForce RTX 2060's with excellent credit/watt and excellent performance.

There were some GTX 1660 Ti's in the scan but not quiet enough to qualify. I'll re-scan in a few weeks and maybe there'll be enough by then.

9) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1970295)
Posted 14 Dec 2018 by Profile Shaggie76
Post:
Hi Shaggie76, I just looked through the github repository and believe there is no reason why you couldn't scan for anonymous hosts running the special app like Juan is always requesting. I see the -anon parameter option listed along with referencing a unique host ID.

If you just passed a known anonymous host ID running the special app through to aggregate.pl that would be a proof of concept I would think. Do I have the correct grasp on the program?
Sure, you can do whatever you want, but like I've said before I am not interested in ranking this app until it's official (and when it is it'll just show up in my charts as the CUDA app like some of the vintage NVIDIA cards already do).

If it doesn't produce the same results as the official build then using it is prioritizing "winning internet points" over "doing science" and I think that's losing sight of the real purpose of this project.
10) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1970196)
Posted 13 Dec 2018 by Profile Shaggie76
Post:
1. I feel like there is a lot of background with how you compile this data. For example, is the credit/hour and credit/watt-hour calculated all-time, or within a timeframe? Do you have a running list of notes somewhere?
It's all open source on github. The credit/hour is based on data crawled from the SETI tasks web pages and the power usage is coarse estimate based on the published TDP for stock cards (I picked the data from the wikiepedia tables).

2. I think I browsed one of your earlier posts, and a different NVIDIA card (the 970)? at one point had a higher credit/watt-hour rating. I'm curious what changed for that.
It depends on the mix of short vs log work units at the time of the scan -- Arecibo work tends to be much faster but as we shift to more Greenbank data the performance characteristics change.

3. Do we know what CPU was used in tandem with the GPU credits? I know the CPU provides a minor role in the crunching of the WU, but I wonder if there is a significant difference between one CPU and another.
I'm not tracking that data but the core script tracks CPU used by the GPU work units if you run it manually. If I recall correctly AMD Open CL and NVIDIA CUDA builds use less CPU than NVIDIA OpenCL builds but this may not matter so much with hyperthreading because even though the GPU tasks's CPU side is busy polling another thread can get work done on the same core because the ALU/FPU units aren't very busy.

Personally prefer 100% GPU where possible because the work/watt is more favorable (and power is the limiting factor for me).
11) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1970134)
Posted 13 Dec 2018 by Profile Shaggie76
Post:
At last I have data:



I'm sorry that the image isn't as sharp as I'd like -- I've transition to a high-DPI laptop and when I convert the graphs to images Excel does weird things. I really should find a better way to finish off the data -- maybe some Perl GD module would be more consistent.

For reference here's the number of hosts and tasks analyzed for the top few cards -- there aren't a lot of 20x0 cards in play yet but they've done enough work that the I feel comfortable with the results .

12) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1970062)
Posted 13 Dec 2018 by Profile Shaggie76
Post:
I finished aggregating the stats but imgur is having technical issues right now and I can't upload it tonight. As you would expect the RTX 2080 Ti steals the performance crown and the 2070 sets a new record for the performance/watt. I'll try to post tomorrow.
13) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1969656)
Posted 9 Dec 2018 by Profile Shaggie76
Post:
Or not -- the servers keep timing out and I can even finish downloading the host database. I'll try to remember in a few days to check again.
14) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1969591)
Posted 9 Dec 2018 by Profile Shaggie76
Post:
I'll try to remember to start a scan before I retire for the night -- it ties up my laptop for a few hours. There might be enough 2080's to qualify by now.
15) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1958711)
Posted 6 Oct 2018 by Profile Shaggie76
Post:
There is host 8587521 which may have been running another card previously 8587521

Nah, that's a new system, created 30/9/2018.
It hasn't been running for even a week, so another 6 or more to go before it's RAC will stabilise (as much as it ever does).

To be clear: my script doesn't use RAC -- it digests all completed tasks. A larger sample size will smooth out averages but after 122 tasks I would expect it to be pretty close.
16) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1958708)
Posted 6 Oct 2018 by Profile Shaggie76
Post:
New this week: there seems to be enough Radeon Pro Vega 64 in circulation now to show up on the charts but sadly it doesn't seem to be any faster than the Vega 56s. The Radeon RX550 edges out the efficiency bracket: maybe the mix of work units or maybe their drivers have been optimized -- who knows?

17) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1958702)
Posted 6 Oct 2018 by Profile Shaggie76
Post:
I did a scan over night and there are about a dozen RTX cards in the results but few have enough OpenCL results yet to get anything but a preliminary picture.

There is host 8587521 which may have been running another card previously -- I don't check the details of each task to avoid overloading the server -- so taken with the largest grain of salt possible here's some preliminary results for a GeForce RTX 2080 (OpenCL).

     955 Credit / Hour
     95% Core / Task
     122 Tasks

I'll have the rest of my usual scan shortly.
18) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1953693)
Posted 4 Sep 2018 by Profile Shaggie76
Post:
Could you post another scan please?

Are you expecting the results to be any different? I don't believe the 2080's have shipped yet.
19) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1945872)
Posted 22 Jul 2018 by Profile Shaggie76
Post:
I collected data for the last two weeks and generated a new set of graphs. Interestingly the average credit/hour for all cards is lower -- maybe we're getting more VLARs these days? My data has long shown that non-vlar Arecibo work units were generally more profitable but it's curious to see how this doesn't seem to favor any particular card. I also noticed that there's much more variance in the Vega data -- I'm wondering if there is ambiguity in the device name returned and as far as the script can tell there's only one type when in fact there's a family of different variants?

20) Message boards : Number crunching : GPU FLOPS: Theory vs Reality (Message 1937251)
Posted 26 May 2018 by Profile Shaggie76
Post:
The thresholds I'm using are at least 10 unique host IDs and at least 25 work units to qualify for each; my results can be combined if I do another scan the next week but I periodically reset to avoid older results with perhaps outdated drivers.

The last scan I posted was a single snapshot and the host counts are here:



Sorry for the image size -- I didn't see how to post tables so I did a hack job of a high-dpi copy from Excel.


Next 20


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.