| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Community
Forum: Hardware Chat Room
Thread: Projects where Ryzen or Intel perform better? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 18
|
|
| Author |
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
Are there any projects that are known to perform better on Ryzen or Intel platform? I recently performed a comparison of my Ryzen 1700 (Ubuntu 17.10) to my i7-3770 (Ubuntu 16.04) running MCM. https://www.worldcommunitygrid.org/forums/wcg...ad,40393_offset,20#555143 I expect that improvement holds for the other projects as well. At least they have not thrown any errors yet. |
||
|
|
wolfman1360
Senior Cruncher Canada Joined: Jan 17, 2016 Post Count: 176 Status: Offline Project Badges: 
|
Are there any projects that are known to perform better on Ryzen or Intel platform? I recently performed a comparison of my Ryzen 1700 (Ubuntu 17.10) to my i7-3770 (Ubuntu 16.04) running MCM. https://www.worldcommunitygrid.org/forums/wcg...ad,40393_offset,20#555143 I expect that improvement holds for the other projects as well. At least they have not thrown any errors yet. Does MCM get impacted negatively when running under windows compared with linux? I'm seeing around 2.5 to 3 hours on my i7-8700 with windows though other things may be using the machine. I've been tempted to make boinc run at a slightly elevated priority on computers I don't use often just to get a little more performance out of them. Good or bad idea? I think there's an option you can edit in the conf file.
Crunching for the betterment of human kind and the canines who will always be our best friends.
AWOU! |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
Does MCM get impacted negatively when running under windows compared with linux? I'm seeing around 2.5 to 3 hours on my i7-8700 with windows though other things may be using the machine. I've been tempted to make boinc run at a slightly elevated priority on computers I don't use often just to get a little more performance out of them. Good or bad idea? I think there's an option you can edit in the conf file. MCM (and MIP) are the better ones under Windows. The last time I checked, it was maybe only a 10% or so reduction as compared to Linux. It is the VINA ones (Zika, OET, FAAH as I recall) that get hammered, along with BEDAM for FAAHV2. I am not so sure about the others, such as SCC and TB, but there are not many of those anyway. My Linux machines are all dedicated, so priority does not matter; BOINC is the only thing that runs (and Folding on the GPUs), so I don't know. [Edit 3 times, last edit by Jim1348 at Dec 12, 2017 6:41:32 AM] |
||
|
|
QuantumEthos
Senior Cruncher Joined: Jul 2, 2011 Post Count: 336 Status: Offline Project Badges:
|
you could buy better fans .... or water cooling!
|
|||
|
||||
|
Macromancer
Veteran Cruncher United States Joined: Sep 6, 2016 Post Count: 994 Status: Offline Project Badges:
|
you could buy better fans .... or water cooling! fwiw, all of my desktop PCs are water cooled (Corsair AIO, single fan radiators). both the Ryzen and I5 PCs are a cool 58C while toiling along at 100% . . . |
||
|
|
Michael Goetz
Cruncher United States Joined: Dec 11, 2017 Post Count: 35 Status: Offline Project Badges: 
|
Projects running the LLR app (PrimeGrid or SRBase), or Prime95 (GIMPS, although that's not a BOINC project), heavily favor Intel CPUs. The AVX and FMA3 implementation on all AMD CPUs to date, including Ryzen, is substantially inferior to Intel's -- at least for the instructions that are heavily used in gwnum which is at the core of both apps.
PrimeGrid's GFN-21 CPU Genefer app also makes heavy use of AVX/FMA3, and like LLR runs much faster on Intel. There's likely other AVX/FMA3 apps out there also run much better on Intel, although I don't know of any that are used in distributed computing projects. |
||
|
|
mmonnin
Advanced Cruncher Joined: Jul 20, 2016 Post Count: 148 Status: Offline Project Badges:
|
Projects running the LLR app (PrimeGrid or SRBase), or Prime95 (GIMPS, although that's not a BOINC project), heavily favor Intel CPUs. The AVX and FMA3 implementation on all AMD CPUs to date, including Ryzen, is substantially inferior to Intel's -- at least for the instructions that are heavily used in gwnum which is at the core of both apps. PrimeGrid's GFN-21 CPU Genefer app also makes heavy use of AVX/FMA3, and like LLR runs much faster on Intel. There's likely other AVX/FMA3 apps out there also run much better on Intel, although I don't know of any that are used in distributed computing projects. Danial (his BOINC ID) has created some custom apps at TN-Grid and Rakesearch that utilize AVX (and higher) functions. My 1950x technically supports AVX2 but when I used that app at Rakesearch it was slower than regular AVX. The Zen die naively only supports AVX and ends up being slower if AVX2 is used.  |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges:
|
There was an unofficial Asteroids@home FMA3 optimised app I tried on an AMD A8-6500 3.5Ghz a few years ago on Windows 64-bit. It ran significantly faster than the official app.
----------------------------------------Optimised apps for WCG sciences would definitely be welcome.  - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz |
||
|
|
|