Message boards :
Number crunching :
Seti@home v8
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 10 · Next
Author | Message |
---|---|
Tom Rinehart Send message Joined: 12 Dec 01 Posts: 113 Credit: 13,255,975 RAC: 6 |
TBar - Are you going to do an SSE41 optimized version of v8 like you did for v7? |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
I've been testing one for a few days, http://setiathome.berkeley.edu/result.php?resultid=4646145743 It's a little slower than the v7 App though, I was hoping to speed it up a little. It's still faster than the stock version on my machine though. |
KC5VDJ - Jim the Enchanter Send message Joined: 17 May 99 Posts: 81 Credit: 4,083,597 RAC: 0 |
is lunatics working on a new enhanced client? Delidded i7-4790K (CLU/CLU) at 4.7GHz @ 1.310Vcore 24/7, 32GB DDR3-2400, Corsair H100i v2, Gigabyte Z97X-Gaming G1 WIFI-BK, MSI Radeon RX 480 Gaming 4G, HX-650 PSU, Corsair 750D |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
is lunatics working on a new enhanced client? We're working on it, but it's not finished yet. |
KC5VDJ - Jim the Enchanter Send message Joined: 17 May 99 Posts: 81 Credit: 4,083,597 RAC: 0 |
is lunatics working on a new enhanced client? Will it take advantage of AVX2 in addition to AVX? In my own benchmarking, it can be up to a 90+% increase in speed. Delidded i7-4790K (CLU/CLU) at 4.7GHz @ 1.310Vcore 24/7, 32GB DDR3-2400, Corsair H100i v2, Gigabyte Z97X-Gaming G1 WIFI-BK, MSI Radeon RX 480 Gaming 4G, HX-650 PSU, Corsair 750D |
Mike Send message Joined: 17 Feb 01 Posts: 34258 Credit: 79,922,639 RAC: 80 |
is lunatics working on a new enhanced client? No, not at this time. AVX just gives a little improvement. With each crime and every kindness we birth our future. |
KC5VDJ - Jim the Enchanter Send message Joined: 17 May 99 Posts: 81 Credit: 4,083,597 RAC: 0 |
is lunatics working on a new enhanced client? A little is better than none, and all modern CPUs since Sandy Bridge have it. Delidded i7-4790K (CLU/CLU) at 4.7GHz @ 1.310Vcore 24/7, 32GB DDR3-2400, Corsair H100i v2, Gigabyte Z97X-Gaming G1 WIFI-BK, MSI Radeon RX 480 Gaming 4G, HX-650 PSU, Corsair 750D |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
is lunatics working on a new enhanced client? I posted the general outline of the plans in the Panic Mode On (101) Server Problems? thread a couple of days ago. It's possible that Eric may catch me out by transferring the apps early in his working day, rather than waiting until I'm on my way to bed - and I won't know until I see them whether the prepared installer is ready to release without modification. We'll have to play it by ear. Yes, I have been provided with AVX versions of the applications, but none of our developers have moved on to AVX2 yet. Not all of our volunteers here can afford to buy the latest hardware on release, so we cater for a wider range of CPUs than just Sandy Bridge - or just Intel, come to that. |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
Of the very few "results returned in the last hour" only 10% are v8 cpu results. It used to be around 80000 for version 7 total ( inluding GPU). The applications site lists about 250,000 GigaFOPS for windows v8. How can it be, that only about 200 results are returned per hour,but lists this enormous gigaflop count? (AP excluded) |
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
No Gpu WUs yet, because no app yet |
tullio Send message Joined: 9 Apr 04 Posts: 8797 Credit: 2,930,782 RAC: 1 |
I see 2000 v8 results/hour, not 200. Tullio |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
That is combined (v7/v8) |
Chris Adamek Send message Joined: 15 May 99 Posts: 251 Credit: 434,772,072 RAC: 236 |
Holy cow, I just saw Petri33's cuda 7.5 result times. I assume this is due to his custom optimizations for the latest (Maxwell and up?) GPU's? That's incredible its that much faster, hopefully we'll all see the fruit of that labor once we get the stock apps out. My 750ti's might start making some noise. lol Chris |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Holy cow, I just saw Petri33's cuda 7.5 result times. I assume this is due to his custom optimizations for the latest (Maxwell and up?) GPU's? That's incredible its that much faster, hopefully we'll all see the fruit of that labor once we get the stock apps out. My 750ti's might start making some noise. lol Most definitely :) A raft of Perti's work to integrate as baseline v8 proves itself for the compatibility/reliability aspects. Some things to nut out about how to allow performance code without sacrifices in other areas, after all I'd like to have my cake and eat it too. What we'll mostly be looking at is how to best to make it run for your specific needs (for example some may have plenty of CPU to feed that kindof throughput). Other's might want to crunch light while playign Xcom long war and youtube videos. Probably will end up like some small gui with a throttle, and some limited automation capability, since deploying apps with fire extinguishers probably isn't practical, lol. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
Holy cow, I just saw Petri33's cuda 7.5 result times. I assume this is due to his custom optimizations for the latest (Maxwell and up?) GPU's? That's incredible its that much faster, hopefully we'll all see the fruit of that labor once we get the stock apps out. My 750ti's might start making some noise. lol Not quite yet - I've just logged an inconclusive against one of his testers with the latest x41p_zi But it was an extremely early overflow (~3 seconds), which we don't worry about too much. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Holy cow, I just saw Petri33's cuda 7.5 result times. I assume this is due to his custom optimizations for the latest (Maxwell and up?) GPU's? That's incredible its that much faster, hopefully we'll all see the fruit of that labor once we get the stock apps out. My 750ti's might start making some noise. lol Yeah, from what I've last seen (a day ago due to dust bunny cleanout time here) it's getting close for high end dedicated crunching, testing many features I haven't had the chance to yet. alpha staging area for the code is in place (along with a raft of slots for x42 work after that). Probably early performance betas will need to be decent systems/newer GPUs etc, though eventually enough bits will fall into place to be flexible enough for stock (optimised releases somewhere in between). "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
tullio Send message Joined: 9 Apr 04 Posts: 8797 Credit: 2,930,782 RAC: 1 |
That is combined (v7/v8) I read the following on the server status page: 2264 V7 145 AP 2264 V8 returned in the last hour Tullio |
Rasputin42 Send message Joined: 25 Jul 08 Posts: 412 Credit: 5,834,661 RAC: 0 |
Do me a favor and read the server page properly. COMBINED! That is why v7 and v8 always read the same. |
tullio Send message Joined: 9 Apr 04 Posts: 8797 Credit: 2,930,782 RAC: 1 |
OK, but how do you now the relative ratio? Tullio |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14653 Credit: 200,643,578 RAC: 874 |
Do me a favor and read the server page properly. Judging by the number of v8 (alone) 'out in the field' and 'waiting for db purging', I really don't trust those "combined" figures - especially turnaround, which is nowhere near a week and a half for v8. I suspect the combined figures are, in reality, v7 only. |
©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.