mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > PrimeNet

Closed Thread
 
Thread Tools
Old 2003-12-12, 00:29   #1
guest
 

23·163 Posts
Default Another reason for high attrition...

After reading through the "M40 again!" thread, it got me wondering what other reasons there could be for the high attrition rate GIMPS sees. Well, I thought of another one that was not mentioned in that thread, which probably accounts for a lot. The high CPU usage was mentioned as a possible reason, and this is related to that, although the high CPU usage itself probably isn't a factor because it doesn't cause lag, being at such low priority.

What it does do, however, is make the processor heat up like mad. Anyone who pays attention to such things (like your average gamer and/or overclocker), will hear alarm bells when Prime95 is running. My own processor (AthlonXP 2200+) die temp increased by 8 degrees during idle times, and is running hotter than I've ever seen it run since I installed Prime95. To be honest, I'm not sure how long I'll stay on due to this one factor.

I think this is probably where you're losing a lot of otherwise extremely receptive hardware geeks. Not much you can do about it, except recommend some cooling solutions to offset the temperature rise.

Anyway, i just thought I'd throw this out there. As I said, there's not much to be done. Cheers.
 
Old 2003-12-12, 01:22   #2
Xyzzy
 
Xyzzy's Avatar
 
Aug 2002

23·1,069 Posts
Default

If you really wanted to run Prime95 and were worried about the heat, you could try the reduced duty cycle option mentioned in this thread:

http://www.mersenneforum.org/showthr...p?threadid=389
Xyzzy is offline  
Old 2003-12-12, 02:10   #3
guest
 

26×43 Posts
Default

Cool, thanks. Although I must say that burying that info in the bottom of a file titled "UNDOC.TXT" is not going to prevent any attrition due to CPU temperature rise. A menu option would probably be better.

I'm not trying to whine, I just don't want to see you lose 80% of the current peak in users, considering that the numbers have doubled over the past month. Especially since with the newest users come the faster processors, one would think.

Speaking of which, are there any stats on the site concerning the search rate? I'd like to get an idea of how increasing processor power is affecting the rate at which new potential primes are being tested. I wonder how the increase in power is offsetting the increase in size of the numbers being tested?

Anyway, Christmas is right around the corner. How many of those 25k+ new users will get a brand new P4 from Santa do you think? :)

Cheers.
 
Old 2003-12-12, 02:15   #4
nfortino
 
nfortino's Avatar
 
Nov 2003

3·5·11 Posts
Default

http://opteron.mersenneforum.org should have most of what you are looking for. Please note that it is no surprise the LL testing rate has not yet increased, as the work is only credited at the end of the test, giving a two week lag. As expected, the trial factoring rate has increased.

Last fiddled with by nfortino on 2003-12-12 at 02:17
nfortino is offline  
Old 2003-12-13, 03:43   #5
cheesehead
 
cheesehead's Avatar
 
"Richard B. Woods"
Aug 2002
Wisconsin USA

22×3×641 Posts
Default Re: Another reason for high attrition...

guest,

Thanks for sharing your thoughts!

Quote:
Originally posted by guest
After reading through the "M40 again!" thread, it got me wondering what other reasons there could be for the high attrition rate GIMPS sees. Well, I thought of another one that was not mentioned in that thread, which probably accounts for a lot. The high CPU usage was mentioned as a possible reason, and this is related to that, although the high CPU usage itself probably isn't a factor because it doesn't cause lag, being at such low priority.

What it does do, however, is make the processor heat up like mad.
The heat problem is very well known here and is mentioned in dozens and dozens of threads (many in the Hardware section) that you may not yet have read. However, your bringing it up points out that there is not a prominent discussion of heat problems in the documentation readily available to beginners.

Quote:
Anyone who pays attention to such things (like your average gamer and/or overclocker), will hear alarm bells when Prime95 is running. My own processor (AthlonXP 2200+) die temp increased by 8 degrees during idle times, and is running hotter than I've ever seen it run since I installed Prime95. To be honest, I'm not sure how long I'll stay on due to this one factor.

I think this is probably where you're losing a lot of otherwise extremely receptive hardware geeks. Not much you can do about it, except recommend some cooling solutions to offset the temperature rise.
... which is plentifully done in this forum's Hardware section, BUT reaching those recommendations requires newcomers (1) to register, (2) to go into the Hardware section and (3) to read enough threads, many of which do not include the word "heat" in their subject lines.

- - - - -

Quote:
Although I must say that burying that info in the bottom of a file titled "UNDOC.TXT" is not going to prevent any attrition due to CPU temperature rise. A menu option would probably be better.
Right.

Quote:
I just don't want to see you lose 80% of the current peak in users, considering that the numbers have doubled over the past month.
Well, some of us figure we'll lose 80% of the current crop of newcomers no matter what, just because of boredom or other factors. :-}

But your point is valid that if we made it easy for newcomers to see effective solutions to the heat problem, we'd retain more.
cheesehead is offline  
Old 2003-12-13, 12:50   #6
dsouza123
 
dsouza123's Avatar
 
Sep 2002

2×331 Posts
Default

Do any other distributed projects mention heat ?

Heat is good, at least in the winter, where I live.

In the summer the extra heat can be an issue and
the suggestion of a menu item such as

Run Cool(er)

would be a nice addition.

I guess in the winter run warm would be fitting.

Most people wouldn't find the information on the
throttling the CPU in the undoc.txt and

In prime.ini add a line
Throttle=x with x some number

and the calculation is really for sleep time and it requires
some trial and error to correlate it with a given temperature.
It probably will still require trying different values but
if it is a menu item it will be easy enough to do.

Many (most) will not know what it means and it's
connection to having the CPU run cooler
or may think it reduces the CPU clock speed (doesn't).

Checking (and cleaning) the heat sink and fans for dust is vital
for good temperature control. I found alot of dust on both on
my PC, after cleaning them off the temperature dropped.

I suggest to only try cleaning with the PC off and in the case of the heat sink, after it has been some time ( an hour ?? ), because the heat sink can get quite hot (gets hot whether prime95 is running or not).
Don't use liquids, just blow the dust out or wipe the finer dust off.
I use an old brush from an electric shaver to clean off the CPU fan and heat sink.

With special thanks to flava for proposing the throttle feature
and to George (Prime95) for implementing it.
dsouza123 is offline  
Old 2003-12-13, 15:49   #7
Quacky
 
Sep 2002

10102 Posts
Default

As a hardware geek myself part of the fun is finding the programs that stress the system the most
Quacky is offline  
Old 2003-12-14, 23:21   #8
QuintLeo
 
QuintLeo's Avatar
 
Oct 2002
Lost in the hills of Iowa

26·7 Posts
Default

Distributed.Net specifically mentions heat somewhere in their docs and/or on their web site.

They're the only other project I've *seen* that generates CPU loading in the same range that Prime95 manages - though they load the Integer unit more, and the FPU not at all, and don't put much stress on memory.

Any "overclocking enthusiast" or "high-end machine" fan worthy of the name (IMO) is already running HD cooling on their CPU - either high-end air cooling along the lines of the Alpha 8045 or Swiftech 462 or more recent high-end offerings, or water cooling.

9-)


As far as the prize money goes - once the first 10,000,000 prime is found, each person in GIMPS that found a lower-size prime gets 5,000 (I think that might be limited on how many people do so, but practically speaking they all will unless we find a Mersenne Prime cluster), George and Scott get some as the developers and maintainers of the GIMPS project, and the largest chunk (probably a majority, depends on how many smaller primes got found first) goes to the person that finds the 10,000,000 digit prime. This IS documented somewhere on the web page, but I forget where offhand.

Last fiddled with by QuintLeo on 2003-12-14 at 23:28
QuintLeo is offline  
Old 2003-12-15, 14:32   #9
flava
 
flava's Avatar
 
Feb 2003

7616 Posts
Default

Quote:
Originally posted by dsouza123


With special thanks to flava for proposing the throttle feature
and to George (Prime95) for implementing it.
You make me blush

I used this feature before the version 23. I just added the Sleep in the source code of Prime95 and recompiled. After the 23 I could no longer compile with my MSDEV6, so I asked George if he can do something about it.

Now I worry less about heat, since the case has good outflow and the CPU is not as hot as it used to be. I noticed something strange, maybe someone can explain it...
I had a Duron1300 with a Taisol cooler (760 or something, it blows towards the CPU). The cooler was mounted right, and the CPU core had enough grease on it (noname brand though). With Prime95 at 100%, in a normal ambient temperature, the CPU was about 56°C (up to 59-60°C in the summer). The MB was at about 40°C. Pretty hot CPU, hugh?
Now I changed my MB and CPU. I have a Athlon XP 2400+ and I kept the same cooler. I got some good grease, put the cooler and tested with Prime95. The CPU is at 49°C, the MB 39°C.
Why this difference? I thought the Athlon XP 2400+ generates a lot more heat than the Duron.
flava is offline  
Old 2003-12-15, 19:54   #10
nomadicus
 
nomadicus's Avatar
 
Jan 2003
North Carolina

2×3×41 Posts
Default

Quote:
Originally posted by flava
Why this difference? I thought the Athlon XP 2400+ generates a lot more heat than the Duron.
Almost the exact same thing happened to me. Switched from an XP1700 to XP2600 and the temp dropped (from 65 down to 46). I think it was the thermal pad.
nomadicus is offline  
Old 2003-12-15, 22:17   #11
guest
 

127F16 Posts
Default

flava, nomadicus:

From some quick calculations I did for my Athlon XP 2200+, the thermal resistance from the die to ambient (ThetaJA) for a 46 degree die temp on a maxed-out part, is around 0.309 degrees C/Watt. For a 54 degree die temp (assuming same power output), this becomes ~0.426 degrees C/Watt. While it is possible that the thermal grease used (or how it's applied) could account for this delta, when you consider that the max power dissipated by a Duron is lower than that of an Athlon, the thermal resistance starts getting a little high for it to be purely an effect of the grease. Likely there is a difference in the ThetaJC value between the Duron and Athlon chip carriers. In any case, when you're working with Theta values so low, a small change in any of a number of factors could have a fairly significant impact.

To all who've responded, thanks for the suggestions/general info. Very useful. After rereading the hardware specifications of my processor, I'm no longer concerned with the generated heat (as long as I can keep it the heck away from my HDDs!). Everything is running smoothly, and Prime just got to the LL stage of testing the exponent I've been assigned. With luck, I'll have a new Mersenne Prime in a little over a month! ;)

Cheers, all.
 
Closed Thread

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Norwegians seems to love primes for some reason kaffikanne Lounge 8 2010-05-27 20:51
The crime of reason jasonp Soap Box 1 2009-12-20 15:09
Is there any good reason for why GMP does this? ShiningArcanine Software 2 2007-12-11 15:11
My fatalistic(sp?) reason for believing in God. jasong Soap Box 22 2006-05-04 19:26
eBay suspending users for no reason? eepiccolo Lounge 12 2003-05-05 19:25

All times are UTC. The time now is 00:52.


Sat Jan 28 00:52:58 UTC 2023 up 162 days, 22:21, 0 users, load averages: 0.51, 0.86, 1.06

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2023, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.

≠ ± ∓ ÷ × · − √ ‰ ⊗ ⊕ ⊖ ⊘ ⊙ ≤ ≥ ≦ ≧ ≨ ≩ ≺ ≻ ≼ ≽ ⊏ ⊐ ⊑ ⊒ ² ³ °
∠ ∟ ° ≅ ~ ‖ ⟂ ⫛
≡ ≜ ≈ ∝ ∞ ≪ ≫ ⌊⌋ ⌈⌉ ∘ ∏ ∐ ∑ ∧ ∨ ∩ ∪ ⨀ ⊕ ⊗ 𝖕 𝖖 𝖗 ⊲ ⊳
∅ ∖ ∁ ↦ ↣ ∩ ∪ ⊆ ⊂ ⊄ ⊊ ⊇ ⊃ ⊅ ⊋ ⊖ ∈ ∉ ∋ ∌ ℕ ℤ ℚ ℝ ℂ ℵ ℶ ℷ ℸ 𝓟
¬ ∨ ∧ ⊕ → ← ⇒ ⇐ ⇔ ∀ ∃ ∄ ∴ ∵ ⊤ ⊥ ⊢ ⊨ ⫤ ⊣ … ⋯ ⋮ ⋰ ⋱
∫ ∬ ∭ ∮ ∯ ∰ ∇ ∆ δ ∂ ℱ ℒ ℓ
𝛢𝛼 𝛣𝛽 𝛤𝛾 𝛥𝛿 𝛦𝜀𝜖 𝛧𝜁 𝛨𝜂 𝛩𝜃𝜗 𝛪𝜄 𝛫𝜅 𝛬𝜆 𝛭𝜇 𝛮𝜈 𝛯𝜉 𝛰𝜊 𝛱𝜋 𝛲𝜌 𝛴𝜎𝜍 𝛵𝜏 𝛶𝜐 𝛷𝜙𝜑 𝛸𝜒 𝛹𝜓 𝛺𝜔