Mersenne Rig (take 2)
 

Hi Mersenners,

About 14 months ago I was in discussion here about getting back into crunching Mersenne primes for the first time since 2001. You can read the old thread here:

[URL]http://www.mersenneforum.org/showthread.php?t=18521[/URL]

Unfortunately, I had some life delays that I won't go into (major health issue) that sort of caused a bunch of havoc. But now I'm back and excited again to build a machine.

The good news is my budget has increased for this project. Can somebody walk me through the best use to spend $1000 on a headless Mersenne cruncher setup? What is the ideal CPU/graphics/RAM? I am ok with buying used equipment if it means more crunches/hour for my budget. I can overclock if you feel it would benefit the situation, however I'm trying to keep the rig somewhat quiet since it'll sit at my feet at my office.

Prime95 suggested this:

[QUOTE]IMO, invest your money in:
a) a K series non-hyperthreaded CPU for overclocking
b) fast RAM, DDR3-2400 or better
c) an efficient power supply, the extra upfront cost should pay for itself in lower electric bills.

IMO, do not invest your money in:
a) motherboard - choose the cheapest board from a reputable maker
b) quantity of RAM
c) disk, case, etc. [/QUOTE]Is this still ideal?

Yes. If you're going modern, the i5-4670K is your best bet. You're not spending the extra $100 on extra threads that serve no purpose. The same applies to older generations. Get the i5-XXXX[B][U]K[/U][/B].

Get faster memory, not higher capacity memory. I'm currently using 125MB on two LL's and two DC's. 32 GB would serve no purpose.

The more efficient power supplies are more important if you're going to run a heavy GPU as well. Where the CPU might draw at most 60W-80W, a GPU can pull 150W-300W depending on what you get.

A PSU that is 80% efficient and requiring 100W draws 125W at the wall. The 25W loss equates to 1 kWh after 40 hours which comes to between 4 and 5 kWh per week, anywhere from $0.50 to $0.75 per week. That's $25 - $40 per year.

A PSU that is 85% efficient requiring 100W draws only 117.6W at the wall, but the 7.4W difference could be $7 - $10 per year. If your system requires 300W instead, triple that amount.


The motherboard need only be powerful enough to support your hardware. I can vouch for ASUS boards being able to give a CPU enough of an overclock to keep 2400 MHz memory bottlenecked. If overclocking, you want to make sure the motherboard is capable of supplying that much power to the CPU, and most boards probably can.


I'll leave the discussion of the "best" GPU to buy to someone else. Used hardware isn't my area of expertise but I know it's the best place to go.

It might be worth waiting for Broadwell in a few months, or longer for DDR4 memory.

I'm not expecting Broadwell to be excitingly much faster, except conceivably on the models with L4 cache. The better Broadwell multiplier might be good for TF, but if you are doing TF on CPUs you have already left the path of wisdom.

For current hardware details I would agree with TheMawn, and would say either to buy now or to wait a year or so for the Sky Lake/K models with 512-bit-wide AVX and large L4 cache to exist and for George to get the Prime95 code to work well on them.

If it fits within budget, go for a 5820K

[url]http://ark.intel.com/compare/80811,75048,82932[/url]

EDIT:- For GPU (TF only), I think 970 is the best one.

Given the price of DDR4 memory, the price of X99 motherboards, and the need for a GPU, I haven't found a 5820K system which is cheaper than two i5-4670K systems.

[QUOTE=fivemack;388636]Given the price of DDR4 memory, the price of X99 motherboards, and the need for a GPU, I haven't found a 5820K system which is cheaper than two i5-4670K systems.[/QUOTE]

I would imagine quite a bit less power usage though.

[QUOTE=henryzz;388642]I would imagine quite a bit less power usage though.[/QUOTE]

Maybe not. At idle, my Haswell system with no GPUs draws 30 watts. The 4790k draws 88 watts at design power, while the 5820k draws 140. At first that looks like 170 watts vs 236 watts. But the two 4790k's give you 8 cores versus the 6 cores of the 5820k, and both consume roughly the same amount on a per-core basis (42.5 vs 44.3). But the 4790k cores are 21% faster at stock clocks for only 4% more power.

[QUOTE=Mark Rose;388644]Maybe not. At idle, my Haswell system with no GPUs draws 30 watts. The 4790k draws 88 watts at design power, while the 5820k draws 140. At first that looks like 170 watts vs 236 watts. But the two 4790k's give you 8 cores versus the 6 cores of the 5820k, and both consume roughly the same amount on a per-core basis (42.5 vs 44.3). But the 4790k cores are 21% faster at stock clocks for only 4% more power.[/QUOTE]

And, as Tom mentioned, the 5820 has no GPU built-in. I think the 5820 only holds its own on tasks where HT is fruitful- e.g. NFS sieving. For pure prime-searching, an i5 is surely the way to go.

I've ordered parts for a Haswell-E system. I sprung for the i7-5930K, as I will likely be adding stuff to it until it is ready to take over as my main computer. I worked a bunch of overtime at my temporary job during fall so anything above a "bare-bones" system is a treat to myself.

As a pure-LL machine, there is absolutely no reason to go for Haswell-E. The only real edge Haswell-E has over Haswell is the memory bandwidth. Where my i5-3570k (admittedly this is only Ivy Bridge) is bottlenecked by 2400 MHz memory around the 4.0 GHz mark, the increased memory bandwidth will likely be able to properly feed six cores at maybe even an increased frequency.

On the other hand, Haswell-E gives you the option to run many GPU's without completely crippling yourself in any applications where you actually need the PCI-E lanes. This is part of my reasoning for spending the extra $200. Even 3 GPU's at x8/x8/x8 only leaves 4 lanes open for whatever PCI-E based devices remain. 40 lanes is certainly overkill, but I'm all for that.

[QUOTE=TheMawn;388671]I40 lanes is certainly overkill, but I'm all for that.[/QUOTE]You can never have too much overkill

GPU crunching?
 

This may be a bit of noob question, but is there any huge reason to get GPU crunching? For pure prime searching I'm feeling an i54670K is all I'm getting into.

[QUOTE=simon389;388762]This may be a bit of noob question, but is there any huge reason to get GPU crunching?[/QUOTE]

No, this project is all about doing what you want to do -- all contributions are welcome. GPUs are superb at trial factoring. This is great for helping the project eliminate candidates and garnering "CPU credit". However, GPUs can be expensive, burn up a lot of electricity and generate a lot of heat. Also, GPU trial factoring will not help you personally discover a new Mersenne prime.

[QUOTE=Prime95;388763]GPUs are superb at trial factoring[/QUOTE]
read not like a bit better, but like 10 to 50 times better, depending on gpu...

Thank the maker for Cyber Monday:

Memory: [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16820233610[/url]
i5-4690k: [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16819117372[/url]
PSU: [url]http://www.newegg.com/Product/Product.aspx?Item=N82E16817438018[/url]

Looking forward to this labor of love.

I missed the Cyber Monday sale price, but even the regular $240 for 2800 MHz isn't exactly terrible. I'll be interested to see how much further it drops. The 3000+ kits are still ridiculous, though.

[QUOTE=TheMawn;388909]I missed the Cyber Monday sale price, but even the regular $240 for 2800 MHz isn't exactly terrible. I'll be interested to see how much further it drops. The 3000+ kits are still ridiculous, though.[/QUOTE]

That's for DDR4, 4x4 DIMMS?

[QUOTE=kladner;388922]That's for DDR4, 4x4 DIMMS?[/QUOTE]

The link in the post above it is for 2x4GB DDR3-2800. 8GB for $240.

I bought 4x4GB DDR4-2666 last week from newegg for $229. Timings are slightly looser, as is the way with DDR4 (CL16 in my case).

Ah! I did not make the connection.

[url]http://www.newegg.ca/Product/Product.aspx?Item=N82E16820231799[/url]

I was fortunate enough to get this kit for $300 a long while back. The rest of the computer is coming shortly and I'm eager to put it to the test.

Alright... we're in the final stages of stress testing. As previously mentioned the chip is 4690K 3.5Ghz. I have gotten pretty solid boots pushing this proc up to 4.7Ghz @ 1.25v, but these collapsed when I ran the small FFT torture test in the latest version of Prime95. I see many overclockers not liking the newest version of the Torture Test since they feel it pushes their temps higher than normal, but since I'm doing distributed computing (and not gaming where there is room for error), I'm wanting a setup that is stable in any test.

Closest I've come so far is 4.3Ghz @ 1.125v on Win7 64-bit with DDR3 @ 2800 12-14-14. Cooled with a Thermaltake Water 3.0 kit. I think I applied the thermal compound a bit thick, so if I finally do get a stable OC I'll probably pull it off and reapply and see if temps go down. Under load it is @ 64C with a few instances of 68C if I run it for an hour. Anything higher than this freezes the computer (BSOD) though interestingly not once has Prime95 gotten any errors. I'm going to let this run all night and see how it fares. I may throw a few other stress test apps at it for good measure.

Thanks for your help in making this happen. I'm excited to rejoin this again.

Simon

Edit: Just connected my v.4 account. Good times.

"not liking the newest version of the Torture Test since they feel it pushes their temps higher than normal"

I am not able rightly to apprehend the kind of confusion of ideas that could provoke such an attitude

[QUOTE=fivemack;390044]"not liking the newest version of the Torture Test since they feel it pushes their temps higher than normal"

I am not able rightly to apprehend the kind of confusion of ideas that could provoke such an attitude[/QUOTE]

Hah I know. Basically, I think the small FFT is the most intense torture test for CPU temps available, and many of them would rather have 95% stable +700mhz overclocks than 100% stable at something a bit lower, like +300mhz. They're attached to the validation the 20% overclock gives them.

In a way, I actually think its ok. After all, most of these hardware boys are really just pushing frames in first person shooters. Their computers don't actually need to be rock solid. But for us doing accurate math distributed computing it seems a very different story.

I got a somewhat stable overclock myself @ 4.3ghz. So I'm now dropping it to 4.2ghz (1.1v @ 66C load) to see if this lasts 36 hours on the Blend torture test.

[QUOTE=fivemack;390044]"not liking the newest version of the Torture Test since they feel it pushes their temps higher than normal"

I am not able rightly to apprehend the kind of confusion of ideas that could provoke such an attitude[/QUOTE]

I think this is hilarious, too. The notion of "stable-for-what" has been lost to most. Some people used to run some heavy photo or video rendering job to stress-test their computer. People used to just play their damn game and see if it crashed. I still see the occasional guy who knows what he's doing suggesting that at the end of the day, you stability test your computer with whatever it is you're going to actually do with it.

Back when overclocking was a thing only for the geeks who are to geeks what geeks are to normal people, some actual effort went into it. Now there's a freaking button you can press and software does it for you! If the overclocking gets easier, all people want is for the stability test to get easier along with it. I'm sure once at least once per week on some of the overclockers' forums there's a new thread created by someone asking what the ultimate stability test software is (without previously looking it up on their own either because they're that kind of person) so that they can hit 'Start' and get a "Stable" or "Not Stable" answer in 30 seconds.

Simon got it pretty much on the dot: there's that sense of entitlement that's got everyone feeling that it's their right to get a 30% overclock and that any less is abuse. Some people complained that Intel's new 8-core monster is a slap in the face because its stock clocks are low and it overclocks poorly. Like Intel just sets stock clocks 1 GHz below normal just to give us the pleasure of setting our processors' "real" power ourselves.

One person at Overclock.net says I chose the wrong processor. He says the ivy bridge would have been better:

"4820k is an enthusiast grade Ivy Bridge-E and there are 6 and 8 core processors that run 12 and 16 threads ."

How is my 4690k better than the 4820k?

Advantages: Lower power consumption; AVX 2 and FMA3, which make mprime/Prime95 run faster; and integrated GPU.

Disadvantages: Less RAM bandwidth per core.

[QUOTE=Mark Rose;391261]Advantages: Lower power consumption; AVX 2 and FMA3, which make mprime/Prime95 run faster; and integrated GPU.

Disadvantages: Less RAM bandwidth per core.[/QUOTE]

But if I had a single computer using Prime95, having the latest 4690k is a faster computer than the fastest 4820k that has 2-4 additional cores?

Basically, for a dedicated Prime95 computer, did I purchase the right processor?

[QUOTE=simon389;391352]But if I had a single computer using Prime95, having the latest 4690k is a faster computer than the fastest 4820k that has 2-4 additional cores?

Basically, for a dedicated Prime95 computer, did I purchase the right processor?[/QUOTE]

Since you can get 2 4690k's for the price of one 4820k, yes. Asking "what does the most work?" needs a price cap, else we can aim you toward some 16-core Xeons for megabucks. If you suffer production envy toward someone's $1200 Haswell-E system, buy a second i5 barebones and know you saved money on two machines vs my one while doing more work too.

[QUOTE=simon389;391352]But if I had a single computer using Prime95, having the latest 4690k is a faster computer than the fastest 4820k that has 2-4 additional cores?

Basically, for a dedicated Prime95 computer, did I purchase the right processor?[/QUOTE]

4820K is a quad-core processor. So is 4690k.
4820K has quad-channel memory. 4690k doesn't.
4820K is 130W TDP. 4690k is 88W.

All of that means that both processors offer similar performance in P95, but 4690K is lot cheaper to buy and run.

[url]http://ark.intel.com/compare/80811,77781[/url]

[QUOTE=VBCurtis;391373]Since you can...[/QUOTE]
:goodposting:
Just to add to it, @simon: which additional cores? for P95 work, logical cores (HT) don't count. P95 is very optimized, taking full advantage of a physical core, you can't use it for a second task with the same priority.

[QUOTE=axn;391390]4820K is a quad-core processor. So is 4690k.
4820K has quad-channel memory. 4690k doesn't.
4820K is 130W TDP. 4690k is 88W.

All of that means that both processors offer similar performance in P95, but 4690K is lot cheaper to buy and run.

[url]http://ark.intel.com/compare/80811,77781[/url][/QUOTE]

Oops- my post referred to Haswell-E, which is the 5820 rather than the 4820. The price-performance tradeoff among 4 core CPUs is heavily tilted toward the fastest i5 available at any given time. You chose well.

Just to follow up on this whole process, I could not get a stable torture test at DDR 2800 at any speed (including stock). I asked the folks at overclock.net for some help, and one of them gave a great piece of advice, which is that these Haswells do not really play well with memory set so high. I dropped it to DDR2400 w/XMP and it was stable.

I then proceeded to run the P95 blend torture test at various overclocks. Despite a watercooled CPU that was at load 60-70C, the best I could get was 4.2Ghz. I then dropped the computer to 4.0Ghz @ 1.5v. It runs at 55C load (once and a while maxes at 59C).

So all of this work just to add a simple 100Mhz over the 4690K's 3.9Ghz and realize this CPU does not like DDR2800 :). No worries.

Anyways, it has reserved a bunch of LL doublechecks and each core is currently doing 50Ghz of work in 6 days. Is this good?

Second question: I set Prime95 to only get first time LL tests, but currently it is on doublechecks. Does this mean P95 needs my computer to prove its stable with doublechecks before it "trusts" me with first time LLs?

Thanks in advance,
Simon

Have you tried DDR3-2666? It would provide a nice boost to throughput.

The DoubleChecks were probably reserved before changing the work type. No matter, as it would be a good idea to run some double-checks until you are thoroughly satisfied that you are stable.

Long-term I recommend going to the web site and selecting the CPUs web page. Click on your new CPU and set the "DC instead of LL percentage". I use 25, but 10 or 15 ought to be good enough. You'll occasionally run DCs and if you get too high a percentage of mismatches, you can take steps to improve stability. I also use that web page to set the option to email whenever the computer submits a suspicious result.

[QUOTE=simon389;392489]
Anyways, it has reserved a bunch of LL doublechecks and each core is currently doing 50Ghz of work in 6 days. Is this good?

Simon[/QUOTE]

I only have 1 close comparison: i5-3570K OC'd to 4.2 is doing about 8/Day/Core of LL; or about 48 per core in 6 days.
It's doing two 56.5M LL per core per month; each worth about 120 GhzDays.
Not offense but I would think the 4690K should do quite a bit better than the 3570K.
Not sure what RAM I'm using off hand; certainly not 2800...might be 2400

Changed the RAM setting from DDR2400 to DDR 2666 and it is still stable. Thanks for that note! The machine is cruising now. Hope the few doublechecks I have lined up don't email me with any errors and we'll be good to go. Thank you everybody for your help.

Simon

Just a four month update... everything is finished and the machine is FAST - cool - silent. I've had a bucket list "life goal" to have a fast Intel computer in the corner of my house crunching Mersenne's and now I have one. Thanks everybody.

How about reliability? Did all DCs checked in successfully?
And just out of curiosity, have you taken any power consumption measurements?

[QUOTE=lycorn;401742]How about reliability? Did all DCs checked in successfully?
And just out of curiosity, have you taken any power consumption measurements?[/QUOTE]

The setup is great! Reliable for sure. Not sure about power consumption. The initial few DC failed, but all after that DCs were successful. I have it set to 10% DCs, so hopefully that is enough to let me know if things should change.

Those 190GhzDay LLs are taking me about 27 days each, give or take.

Below you can see my results.

Username: LendingMemo
Computer is called the_beast. Failed DCs are in [B][COLOR="Red"]red[/COLOR][/B]. Successful DCs are in [B]black[/B]. LLs are [B][COLOR="Blue"]blue[/COLOR][/B].

[COLOR="Blue"]the_beast 71436263 C - Unverified 2015-05-08 19:09 27.5 ACCE6FCB10F3C9__ 190.9928
the_beast 71401007 C - Unverified 2015-05-08 11:04 27.7 C8F3F2EE4CC6AC__ 190.8985
the_beast 66714079 C - Unverified 2015-05-07 04:36 24.9 4D0C9FCCF6C22E__ 168.6384
the_beast 66564317 C - Unverified 2015-04-21 01:32 24.5 7F77D3EF596E46__ 168.2598[/COLOR]
[COLOR="Black"]the_beast 66581051 C - Verified 2015-04-17 01:05 24.2 0618FC3ACF53C7__ 168.3021[/COLOR]
[COLOR="Blue"]the_beast 66572321 C - Unverified 2015-04-16 03:13 24.1 016DD683750BDD__ 168.2800
the_beast 66569221 C - Unverified 2015-04-15 17:20 24.2 F8EA5F276A5CEA__ 168.2722
the_beast 65385709 C - Unverified 2015-04-01 08:43 22.5 1162903E51C50D__ 158.6058
the_beast 66977753 C - Unverified 2015-03-28 12:44 24.8 733259D464720E__ 169.3049
the_beast 66498001 C - Unverified 2015-03-27 17:37 24.5 98C7F744FC1EA5__ 168.0922
the_beast 66473971 C - Unverified 2015-03-27 05:50 24.5 79963FA40BCCC2__ 168.0314[/COLOR]
[COLOR="Black"]the_beast 34533847 C - Verified 2015-03-13 10:19 10.3 8C394227C216E6__ 40.8651
the_beast 34506673 C - Verified 2015-03-09 19:29 11.3 903C2913954B8D__ 40.8329
the_beast 34506149 C - Verified 2015-03-09 19:29 11.7 D003C13DA1FEB4__ 40.8323
the_beast 34505599 C - Verified 2015-03-09 19:29 11.8 90E106319D4869__ 40.8316
the_beast 34502161 C - Verified 2015-03-07 20:14 10.1 17C240ED1E4F1B__ 40.8276
the_beast 34482319 C - Verified 2015-03-03 20:26 10.3 E5151760B63924__ 40.8041
the_beast 34481329 C - Verified 2015-03-03 09:02 10.4 D58942E48ECB45__ 40.8029
the_beast 34480997 C - Verified 2015-03-03 07:26 10.4 AC509FDA722A34__ 40.8025
the_beast 34480543 C - Verified 2015-03-02 20:32 10.1 234B0DA7FE882B__ 40.8020
the_beast 34453063 C - Verified 2015-02-26 18:43 10.2 116C733F4F56DE__ 40.7695
the_beast 34449491 C - Verified 2015-02-26 07:30 10.4 5BFA970FF78DB5__ 40.7652
the_beast 34448857 C - Verified 2015-02-26 05:49 10.5 59FE71A45CC2BC__ 40.7645
the_beast 34448879 C - Verified 2015-02-25 20:34 10.1 F52993EAEA4EB1__ 40.7645
the_beast 34424987 C - Verified 2015-02-21 17:06 10.0 46C467477D0054__ 40.7362
the_beast 34422643 C - Verified 2015-02-21 05:54 10.0 1869FA3723F4F8__ 40.7335
the_beast 34422613 C - Verified 2015-02-21 04:19 9.9 36885CD74CC5BC__ 40.7334
the_beast 34422077 C - Verified 2015-02-20 21:02 9.8 30662FA4F541E3__ 40.7328[/COLOR]
[COLOR="Blue"]the_beast 66048943 C - Unverified 2015-02-16 15:23 24.4 AED1295FCBFFD4__ 160.2146
the_beast 66044807 C - Unverified 2015-02-16 04:56 24.2 3C1DC86CB8849A__ 160.2045
the_beast 66044789 C - Unverified 2015-02-16 02:48 24.1 59AD553AB8214D__ 160.2045
the_beast 66044801 C - Unverified 2015-02-15 21:37 23.9 3FC48542A5E2E0__ 160.2045[/COLOR]
[COLOR="Black"]the_beast 37024159 C - Verified 2015-01-28 12:42 12.5 20322588E548C0__ 46.2802
the_beast 37024177 C - Verified 2015-01-28 05:45 12.2 17011F93025060__ 46.2802[/COLOR]
[COLOR="Red"]the_beast [COLOR="Red"]37024129[/COLOR] C - Unverified 2015-01-28 04:10 12.1 88B84703CDC03B__ 46.2802[/COLOR]
[COLOR="Black"]the_beast 37024063 C - Verified 2015-01-28 04:08 12.1 B542AF54458224__ 46.2801
the_beast 37307533 C - Verified 2015-01-22 15:00 17.4 91A0AFC52FFDF6__ 49.1216
the_beast 37307551 C - Verified 2015-01-22 10:05 17.2 CF4E614791E584__ 49.1216
the_beast 37307489 C - Verified 2015-01-22 09:54 17.1 FBCC0C82F5C586__ 49.1215
the_beast 37307471 C - Verified 2015-01-22 09:04 17.1 71DAA0DD751AD8__ 49.1215
the_beast 39288647 C - Verified 2015-01-15 09:46 10.6 F3D8E02C7ECDA0__ 51.7301
the_beast 39288583 C - Verified 2015-01-15 08:08 10.6 0B7EDDCC3081BB__ 51.7300
the_beast 39288553 C - Verified 2015-01-15 07:19 10.5 8DF5D882F603F8__ 51.7299[/COLOR]
[COLOR="Red"]the_beast [COLOR="red"]39288653[/COLOR] C - Unverified 2015-01-15 07:11 10.5 5755D6A5F0498D__ 51.7301[/COLOR]

For the two red results you [B]don't know yet[/B] that the result is bad. It may be the original LL being wrong. [edit: It happens to me all sometime, and when I triple-check, my residue is right.]

[QUOTE=LaurV;402165]For the two red results you [B]don't know yet[/B] that the result is bad. It may be the original LL being wrong. [edit: It happens to me all sometime, and when I triple-check, my residue is right.][/QUOTE]

Probably true, especially since the other runs verified okay.

[QUOTE]the_beast 37024129 C - Unverified 2015-01-28 04:10 12.1 88B84703CDC03B__ 46.2802
the_beast 39288653 C - Unverified 2015-01-15 07:11 10.5 5755D6A5F0498D__ 51.7301[/QUOTE]

Tell ya what, I'll pick up those 2 mismatches and have an answer for you by tomorrow. (CORRECTION: In 15 and 17 hours respectively)

[QUOTE=Madpoo;402195]Probably true, especially since the other runs verified okay.



Tell ya what, I'll pick up those 2 mismatches and have an answer for you by tomorrow. (CORRECTION: In 15 and 17 hours respectively)[/QUOTE]

So cool! Thanks!

Is 27 days for a 190GhzDay LL exceptionally fast, or is it just par for the i5-4690K course?

[QUOTE=simon389;402221]So cool! Thanks!

Is 27 days for a 190GhzDay LL exceptionally fast, or is it just par for the i5-4690K course?[/QUOTE]

It'll depend on the exact FFT size it ran at... you might want to run the benchmark in Prime95 and compare with others who have the same i5-4690K:
[URL="http://www.mersenne.org/report_benchmarks/?exp_date=2015-01-01&64bit=1&exover=1&exbad=1&exv25=1&exv26=1&specific_cpu=4377405"]Benchmarks for i5-4690K[/URL] (I limited that to just benchmarks since Jan 1 of this year).

You can look at the 4MB FFT size timings for an idea of what you could expect on your own system. Those times are in milliseconds per iteration. So the total time (like 27 days) would depend on the exact exponent you were dealing with... it's better to measure the per iteration times instead.

[QUOTE=simon389;402221]Is 27 days for a 190GhzDay LL exceptionally fast, or is it just par for the i5-4690K course?[/QUOTE]

It isn't. 190GHz Day / 27 days = 7 GHday/day. According to [url]http://www.mersenne.ca/throughput.php?cpu1=Intel%28R%29+Core%28TM%29+i5-4690S+CPU+%40+3.20GHz|256|6144&mhz1=3500[/url], your CPU should be getting about 13-14 GHDay/day. So this is about half of that. Something's off.

[QUOTE=axn;402226]It isn't. 190GHz Day / 27 days = 7 GHday/day. According to [URL]http://www.mersenne.ca/throughput.php?cpu1=Intel%28R%29+Core%28TM%29+i5-4690S+CPU+%40+3.20GHz|256|6144&mhz1=3500[/URL], your CPU should be getting about 13-14 GHDay/day. So this is about half of that. Something's off.[/QUOTE]
Memory in the wrong slots, so running single-channel instead of dual-channel?

[QUOTE=Antonio;402232]Memory in the wrong slots, so running single-channel instead of dual-channel?[/QUOTE]

That would be my first guess.

[QUOTE=Madpoo;402195]Tell ya what, I'll pick up those 2 mismatches and have an answer for you by tomorrow. (CORRECTION: In 15 and 17 hours respectively)[/QUOTE]

They finished overnight, just remembered to check. Congrats, you win on both of those. :smile:
[URL="http://www.mersenne.org/M37024129"]M37024129[/URL]
[URL="http://www.mersenne.org/M39288653"]M39288653[/URL]

[QUOTE=Madpoo;402246]They finished overnight, just remembered to check. Congrats, you win on both of those. :smile:
[URL="http://www.mersenne.org/M37024129"]M37024129[/URL]
[URL="http://www.mersenne.org/M39288653"]M39288653[/URL][/QUOTE]

Awesome! Thanks.