• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Rakesearch stress test for possible DC-Vault inclusion

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Is RakeSearch stable enough of a project during Stress Testing????


  • Total voters
    6
  • Poll closed .
Okay, I have some cores on it now. Don't have time to see if the right/best app is being automatically chosen, though. Might do that later if I have time. I could only skim the thread, is there something in particular that can be looked for and changed to produce better results? I have a motley selection of CPUs from Sandy Bridge through Coffeelake, also two Ryzens...
 
Okay, I have some cores on it now. Don't have time to see if the right/best app is being automatically chosen, though. Might do that later if I have time. I could only skim the thread, is there something in particular that can be looked for and changed to produce better results? I have a motley selection of CPUs from Sandy Bridge through Coffeelake, also two Ryzens...

zzuupp post the link for the optimized apps along with app_info.xml.

https://github.com/sirzooro/RakeSearch/releases/tag/v1.0

On my 3930K running linux, the avx optimized app is at least 10x faster than the "stock" app.
 
Forum posts which describe the optimized applications:
http://rake.boincfast.ru/rakesearch/forum_thread.php?id=39

Which of the applications to install:
  • avx512: Skylake-X, -SP
  • avx2: Haswell and later Intels
  • avx2nopext: Ryzen and later AMDs
  • avx: Sandy Bridge and Ivy Bridge
  • sse2: older Intels
Correct me if I misunderstood.
Not sure about AMDs before Ryzen.

Edit,
in order to verify that one application version is faster than another on your particular host, use the test WU which is provided in "test.tgz" at the github link.
 
Last edited:
@StefanR5R , I know that I can't spend the time to test apps on 10 hosts. It's probably safe to use an app for an older CPU on newer one, I would think? I wonder if there any notable exceptions to the list you posted above?
 
It's probably safe to use an app for an older CPU on newer one, I would think?
The rakesearch forum thread shows this performance on Haswell:

avx2 ................. 100 %
avx2nopext ....... 96 %
avx ..................... 94 %
sse2 ................... 90 %
stock ................. 10 %​

One user tested the avx2 and avx512 versions on Skylake-X, and they perform the same (except when the processor is overclocked such that the AVX512 clock offset is eliminated). Those measurements were done with just one process at a time though, not with #processes = #cores or #threads. This means that potential cache or memory bottlenecks were not covered in this user feedback.

The entire thread lacks feedback about performance of the application versions on the various AMD CPUs.
 
I feel a bit sheepish for asking others to do my homework for me, but I'm a bit overwhelmed today. Thanks, guys, I really appreciate it! I have managed to get most of my hosts on the modded app.
 
😉

@[H]Coleslaw....what is a passing grade on this test? So far, it's been pretty hard to access the webpage/attach a new client. The 7 day run time a test to see if they upgrade/fix the problem? They are losing points though, in my book.

Well, the grade is basically whether people post significant issues with the project or not. If people think it is a turd, then we won't add it. If they think it is acceptable, then we will. It is best to post final reviews of the event in the DC-Vault forums though so that there is more of a community discussion to it from various sides. I personally have not been running it on this challenge as I've gotten my fill of the project. It runs solid year round and has work consistently. I just don't know if it can handle multiple teams hammering away at it or not. I don't care if the site is responsive in general but rather if work is coming and going as needed. But that is just my personal thought on the matter. Ultimately it comes down to what the public bring to the table and then Rusty puts it into effect. I'm just a single voter.
 
It appears 10 has brought the "kitchen sink" on this stress test!!! WHOA BABY!!!! You could have at the very least given everyone a 5 day head start before you brought the hurt!!! 🙂
 
The avx2 application version triggers the AVX clock offset on Kabylake (if there is one configured in the BIOS; most BIOSes don't have one by default), but not on Broadwell-EP. BDW-EP run it at the non-AVX turbo clock.
 
Well, the grade is basically whether people post significant issues with the project or not. If people think it is a turd, then we won't add it. If they think it is acceptable, then we will. It is best to post final reviews of the event in the DC-Vault forums though so that there is more of a community discussion to it from various sides. I personally have not been running it on this challenge as I've gotten my fill of the project. It runs solid year round and has work consistently. I just don't know if it can handle multiple teams hammering away at it or not. I don't care if the site is responsive in general but rather if work is coming and going as needed. But that is just my personal thought on the matter. Ultimately it comes down to what the public bring to the table and then Rusty puts it into effect. I'm just a single voter.

I give a thumbs up. Other TeAm members, please vote (if I can change the thread).
 
Thanks for the input.
As I see it the TeAm wants to vote for the following projects:

Rosetta@Home
CSG
Asteroids@Home
Seti@Home
 
TeAm members, please vote
I am missing the "Meh. Nice badges though" option.

I was just awarded the badger badge.

Badge_Badger.png
 
Back
Top