• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

PrimeGrid Races 2018

Page 25 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Howdy

Senior member
Nov 12, 2017
569
465
106
Oh POOP! I had my days mixed up.

I'm a bit later than usual, but I'm in now.
Apparently it's a "thing" the last few weeks to be either a day early OR a day late.

EDIT:
Thursday last week.......
Me: See you Monday, have a good weekend!!
Boss: Where you going tomorrow? It doesn't show you off on the schedule.
Me: Why do you want me to come in tomorrow? A Saturday, OT, no parts and nothing else to do. Is something coming in?
Boss: Umm tomorrow is Friday............and your parts will be here early AM.
Me: Ummmm.....well then, Have a good night see you tomorrow!!!
 
Last edited:
  • Like
Reactions: zzuupp

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
15,526
2,585
55
It's hard to remember what time of day to post stats, so I'm going to be fairly random, I think.

Day 2 stats:

Rank___Credits____Username
16_____5238618____xii5ku
19_____4192005____Howdy2u2
46_____2094663____iwajabitw
67_____1347345____emoga
91_____1045214____Ken_g6 :D
101____897230___10esseeTony

Rank__Credits____Team
4_____40385186___Czech National Team
5_____24544164___Crunching@EVGA
6_____17523228___AMD Users
7_____14815078___TeAm AnandTech
8_____12557694___Rechenkraft.net
9_____11660822___BOINC@MIXI
10____11069651___Storm

Well, I'm making good progress. Let's all try to pass @TennesseeTony! ;)
 
  • Like
Reactions: StefanR5R

zzuupp

Lifer
Jul 6, 2008
14,850
2,318
126
It's hard to remember what time of day to post stats, so I'm going to be fairly random, I think.

Day 2 stats:

...

Well, I'm making good progress. Let's all try to pass @TennesseeTony! ;)
I didn't make day 2 stats, but I now have 1 finished, 2 on the way.
78K seconds on the red card. The older green cards are going to take much longer.
 

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
Looking at other forums, it occurs to me that many of our competitors are bound to increase their productivity for several reasons:
  • lots of Windows users fighting with system updates during the first one or two days; I suppose it won't trouble them for the rest of the race
  • many people using CPUs along with GPUs, and therefore
    • taking a longer while before their first CPUs tasks are done,
    • still figuring out which # of tasks/processor and # threads/task to set,
    • overcoming trouble with properly configuring a single client for CPU tasks and GPU tasks
I am running only GPUs myself, and not all of them, but am still putting almost 2 kW into this. Nevertheless this may not be sufficient to remain in the top 20, with ~7½ days still to go.
 

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
15,526
2,585
55
Better Nate than Lever, Day Pi stats: ;)

Rank___Credits____Username
17_____8387551____xii5ku
21_____7491748____Howdy2u2
44_____3893055____iwajabitw
83_____2096833____emoga
93_____1646789___10esseeTony :)
103____1493571____Ken_g6 :(
202____448589_____zzuupp
248____149918_____SlangNRox

Rank__Credits____Team
4_____74524136___Czech National Team
5_____41487011___Crunching@EVGA
6_____26223543___AMD Users
7_____25608058___TeAm AnandTech
8_____21690865___Rechenkraft.net
9_____20198573___BOINC@MIXI
10____19163142___Storm

We are so close to passing AMD!
 

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
Has anyone ran the CPU side of GFN21s?
PG's own challenge thread has some reports on GFN-21 on CPUs.
Especially message 123194 by mackerel and message 123219 by Yves Gallot.
mackerel said:
6700k, 24h, 68W
5930k, 25h, 117W
8086k, 26h, 76W
[Ryzen] 2600, 30h, 61W average, 75W peak
...
For comparison, my RTX 2070 does 3 units in just over 24h, and takes reported 150W constantly. That puts it a fair bit more power efficient than CPU.
Yves Gallot said:
Intel(R) Core(TM) i9-7980XE CPU @ 2.60GHz - 17 threads
27,416 sec = 7 h 36 mn 56 sec
Also, message 123470:
Yves Gallot said:
There is a memory bottleneck with 9th generation i7s and i9s.
L3 cache size of i7-9700K is 12 MB and cpuGFN21 memory size is 16 MB R/W + 8 MB Ro.
8 threads run at about the same speed than 4.
In other words:
Desktop CPUs have less cache than needed to fit a single task, and should run at most one cpuGFN21 task at once.​
Desktop CPUs with dual channel memory and Skylake based cores scale only up to 4 threads for a cpuGFN21 task.​
The aforementioned i9-7980XE has 24.75 MB shared L3 cache (non-inclusive, together with 1 MB L2 cache for each of the 18 cores). Evidently this cache setup supports cpuGFN21's demands pretty well.

Some data reported in SETI.Germany's PG challenge thread:

Coffeelake(?) 6-core i5-8600K @4.3 GHz running one cpuGFN21 task along with one GPU task:​
6 cores: 32 h​
5 cores: 30 h​
4 cores: 29.5 h​
3 cores: 32.5 h​
2 cores: 42 h​
If more than one CPU task is run at a time, overall throughput goes down.​
Ryzen 2700X@4.1 GHz, 3466 RAM:​
2 tasks at once, 4 threads per task: 70 h per task​
1 task at once, 8 threads per task: 29 h per task​
Ryzen 1300X and Core i5 4460 both take 38 h per task.​

Based on Yves Gallot's post regarding the memory size of CPU tasks, I now started 2 tasks per processor on
Xeon E5-2696 v4 with 55 MB cache and 22C/44T @ 2.6 GHz in dual-processor machines running Linux.
2 tasks per processor, 11 threads per task:​
15...19 h estimated run time per task after >1.5 % completion
14.5...15 h estimated run time per task after 10 % completion​
2 tasks per processor, 20 threads per task:​
run time estimation hasn't stabilized yet, but "top" shows ~2000 % processor utilization per task, which indicates good scaling.
16.1...18 h estimated run time per task after 10 % completion​

TDP of these processors is 150 W. But since my particular mainboard & BIOS drives them at the all-core avx turbo clock instead of the base clock, it is possible that they persistently consume more than the TDP. Still, with an estimated ~7.5 h mean time between task completions per processor, these 14 nm CPUs are both faster and more efficient than my 16 nm GPUs (1080Ti with ~8.9 h per task at 250 W GPU power + a small amount of power for the supporting CPU).

Xeon E5-2690 v4 with 35 MB cache and 14C/28T @ 2.9 GHz, ditto dual-processor machines running Linux:
1 task per processor, 14 threads per task:​
10.1 h estimated run time per task after 10 % completion​
These processors have 135 TDP, but again, since the particular motherboard drives them at their all-core avx turbo clock instead of base clock, actual sustained power consumption possibly exceeds TDP.

Edit: run time estimation updated
Edit 2: E5-2690 v4 tested too
Edit 3: Background info and more CPU data from prerelease versions and lower "leading edge" can be found in the PG forum thread "genefer 3.3.4".
 
Last edited:
  • Like
Reactions: TennesseeTony

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
AMD Users had some kind of downtime:
I believe you should have placed that in a "Spoiler" :(
Well, here is another post with a lack of spoiler tags:
Seeing how we caught up with AMD Users, I now added 2 more GPUs (to the detriment of F@H, sadly) and 5 of those nice Xeons (somewhat reducing my WCG output, and stopping my activity at some other projects).
 

SlangNRox

Golden Member
Oct 9, 1999
1,010
1
76
Not that I plan on completing a lot on my gtx 1050ti, but it looks like the first wu I started 3hrs after the start of the race wasn't counted. No cpu tasks for me.
 

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
I am seeing a GFN21 task in your results list which was downloaded 3h 15m before the start of the challenge (and therefore it doesn't count in the challenge, unfortunately), and another completed task which was downloaded 32 h after the start (which counts, and puts you at rank 258 of 322 currently).
 

Howdy

Senior member
Nov 12, 2017
569
465
106
Well, here is another post with a lack of spoiler tags:
Seeing how we caught up with AMD Users, I now added 2 more GPUs (to the detriment of F@H, sadly) and 5 of those nice Xeons (somewhat reducing my WCG output, and stopping my activity at some other projects).
Well, not necessarily a spoiler. I guess the move means AMD is in the crosshairs.
 

TennesseeTony

Elite Member
Aug 2, 2003
4,127
3,433
136
www.google.com
Thanks Stefan!

I had a panic attack when I read this post from Michael Goetz "46% came back with some kind of an error" , but my ancient AMD cards have no errors or invalids to date, and they even have a 45% 1st! rate. :)

I sure hope all those errors are the nearly instant type, and not the kind that run for hours upon hours and then gives the error. EDIT: They are mostly instant for the GPU, likely caused by Windows10 over-riding existing drivers which remove OpenCL support and the user not yet noticing the change. And the CPU errors were 75% aborted tasks. So that is good to know the tasks aren't at fault.

Code:
After day 3 of 10:


Challenge: Winter Solstice
App: 16 (GFN-21)
(As of 2018-12-14 22:28:21 UTC)

18376 tasks have been sent out. [CPU/GPU/anonymous_platform: 5619 (31%) / 12757 (69%) / 0 (0%)]

Of those tasks that have been sent out:

8497 (46%) came back with some kind of an error. [2178 (12%) / 6319 (34%) / 0 (0%)]
4004 (22%) have returned a successful result. [574 (3%) / 3430 (19%) / 0 (0%)]
5875 (32%) are still in progress. [2867 (16%) / 3008 (16%) / 0 (0%)]

Of the tasks that have been returned successfully:

2183 (55%) are pending validation. [354 (9%) / 1829 (46%) / 0 (0%)]
1808 (45%) have been successfully validated. [220 (5%) / 1588 (40%) / 0 (0%)]
1 (0%) were invalid. [0 (0%) / 1 (0%) / 0 (0%)]
12 (0%) are inconclusive. [0 (0%) / 12 (0%) / 0 (0%)]

The current leading edge (i.e., latest work unit for which work has actually been sent out to a host) is b=333264. The leading edge was at b=308008 at the beginning of the challenge. Since the challenge started, the leading edge has advanced 8.20% as much as it had prior to the challenge!



So far, two GFN-18s have been found during the challenge!?!

Can we find a GFN-21?
 
Last edited:

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
Competing against "AMD Users" may remain a pointless pursuit. During the past 3 hours, their relevant contributor busied himself adding a whole bunch of new Tesla V100-SXM2 virtual machines at Amazon Web Services.

Though I suspect this is merely a reaction to failing to be top individual user. User #1 on the other hand, who is another user of virtualized V100-SXM2 and P100 servers (at Google Cloud Platform) along with CPU-only servers (perhaps cloud VMs too), may quite possibly even be unaware of this challenge.
 

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
15,526
2,585
55
Day 4 stats, approximately on time!

Rank___Credits____Username
18_____10790855___xii5ku
21_____9894859____Howdy2u2
48_____4794303____iwajabitw
74_____2998129____emoga
95_____2097268___10esseeTony
103____1942240____Ken_g6
200____598459_____zzuupp
266____149918_____SlangNRox

Rank__Credits____Team
3_____98502999___Czech National Team
4_____96722604___SETI.Germany
5_____52593441___Crunching@EVGA
6_____33266034___TeAm AnandTech
7_____31479610___AMD Users
8_____29187029___Rechenkraft.net
9_____26500399___BOINC@MIXI

Well, at least for the moment, we're ahead of AMD. :)
 

SlangNRox

Golden Member
Oct 9, 1999
1,010
1
76
Doh! Looks like I'll probably only be good for 4 wu during this race

I am seeing a GFN21 task in your results list which was downloaded 3h 15m before the start of the challenge (and therefore it doesn't count in the challenge, unfortunately), and another completed task which was downloaded 32 h after the start (which counts, and puts you at rank 258 of 322 currently).
 

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
Competing against "AMD Users" may remain a pointless pursuit. During the past 3 hours, their relevant contributor busied himself adding a whole bunch of new Tesla V100-SXM2 virtual machines at Amazon Web Services.
He continued to do so and has created ~100 instances with 1 task in progress each since yesterday. He hasn't uploaded any of those yet, even though most of them should be complete by now. In addition, he has ~150 previously created V100-SXM2 instances, each of them with 1 task in progress as well. So, his bunker is approaching ~40 M credits now if these cloud instances are still in his control.
 
Last edited:

Ken g6

Programming Moderator, Elite Member
Moderator
Dec 11, 1999
15,526
2,585
55
Strange Day 5 stats:

Rank___Credits____Username
12_____19058687___xii5ku
22_____13201158___Howdy2u2
52_____5846484____iwajabitw
62_____4501159____emoga
84_____3448862____Howdy2u2
108____2391206____Ken_g6
204____748681_____zzuupp
258____300058_____SlangNRox
301____150073_____IEC

Rank__Credits____Team
3_____132579716___SETI.Germany
4_____131799113___Czech National Team
5_____65513565___Crunching@EVGA
6_____49646372___TeAm AnandTech
7_____39537532___Rechenkraft.net
8_____37042946___AMD Users
9_____35209948___BOINC@MIXI

@TennesseeTony seems to have disappeared.
Howdy do that? ;) (It took me about 10 minutes to figure it out!)

Looks like we have to watch out for a (w)reched craft behind us.
 

Howdy

Senior member
Nov 12, 2017
569
465
106
Well it certainly took me by surprise!!! Scrolling down and seeing where everyone is and o_O there's another me!!! That should scare the hell out of anyone!!!

Thanks for the stats Ken!!
 

StefanR5R

Elite Member
Dec 10, 2016
4,285
4,868
136
It's rare for the organizers to comment on the scoreboard.
Michael Goetz said:
On the leaderboards, there's been quite the battle raging between the #1 and #2 team positions as well as the #3 and #4 team positions. They keep flipping back and forth. With tasks worth 150K points each, a million point lead can vanish in a heartbeat.

On the individual side, Ryan Propper has a decent lead in first, but #2, #3, and #4 are very close together.

With half the challenge to go, and with server-class Tesla V100-SXM2 GPUs available to rent that can run circles around an RTX 2080 Ti, completing GFN-21 tasks in only 3 hours and 20 minutes, nothing is close to being decided yet.
Here is a comparison between AWS EC2 prices (regular rates, not spot prices) vs. operational cost in a German household (i.e. with higher electricity costs than in many other countries):
  • An AWS p3.2xlarge instance (one Tesla V100) costs $3.823 per hour in region "EU (Frankfurt)". This is ~$13 per GFN21 task. Note, I never dealt with cloud computing, and didn't browse the AWS pages much, so I may have missed something important in this quick "research".
  • A GTX 1080Ti at stock settings draws 250 W when running GFN21. Together with supporting CPU, the rest of the system, and PSU inefficiency, somewhat more than 300 W are required per 1080Ti. Its GFN21 task duration is 8.9 hours. At a German household electricity rate of a little less than 0.3 €/kWh, this works out to ~0.8 € per GFN21 task for electricity alone. Add a small amount for the initial investment into the DC rig, and other operational cost (such as you having to remove dust once in a while, to mess with the operating system; having the rig taking up space in your home, etc.).
So, outside of using some leftover cloud computing hours from a moneymaking project, or using certain gratis offers on much slower VMs, I don't see cloud computing to be attractive to a regular Distributed Computing hobbyist. Am I missing something, or does one really have to be really desperately wanting to win a DC challenge to go and rent cloud computers for it?
 

Howdy

Senior member
Nov 12, 2017
569
465
106
@StefanR5R,
I am glad you did the calculations for this (well I was hoping you would anyway) I looked this up myself and not knowing exactly how to calculate it, came up with the same question...........WHY? The only reason I could think of was finding a GFN21.....which would be a world record? Then again, why only during a challenge if that was so important?
 

ASK THE COMMUNITY