i7 920 vs i7 860 direct test comparison

[Otis11]

I'll think about it. I have a day job you know.....

You mean you actually have one of those?

I've heard about them... but ... I mean... What' it like?

[123bob]

Huh? Naw, 18x uncore should work with 8x dram... there is no 9x dram divider, next would be 10x which wouldn't work with 18x uncore (for Bloomfields, uncore must be 2x dram speed or higher).

Try higher VTT and Vdimm for higher uncore.

Edit: Ah, you mean the board only shows 9x for a real uncore mult of 18x? Weird.. but then again, it's MSI, so anything is possible
See if it has any "preview" option for uncore speed, 16x is 2960, 18x should be 3330 at your BCLK of 185

Something weird here....Your right, the lowest multi in the bios is 10x. doing that shows 3.7 gig for uncore in the preview....

Something AFU on the math in the bios?. It shows in 3.7 in CPUz too...I can't seem to match the 860. Is it a function of dual vs triple maybe?

I'll have to pick up on this later. Time to work the day job.

Bob

[jcool]

Lol, no.. lowest uncore mult should be 12x, probably 6x in your bios then. Maybe a bos update?

Edit: If you want, I could match a 920 or w3520 to your 860. Would just need those test WUs ^^

[123bob]

Bios update in progress. I am 2 versions behind. The notes on both newer version said "memory compatibility" changes. Isn't it wonderful how detailed that message is????

I'll report back in a few. Hope I don't blow the rig into the dirt....

It is weird that I set uncore multi to 10x and it shows 3.7 gig, exactly the same as the dang CPU at 20x.....

If needed, I can send you the WUs.

Bob

[jcool]

Not weird at all bob, if the board is really doubling all the mults - 185*20 is 3700Mhz, after all
What it should do, however, is let you change uncore from 12x to 40x or so. 12x being the slowest at 12x185=2220Mhz.

On the LGA1156 CPUs, the uncore mult is locked. For the 860 it's locked to 18x, for the smaller ones like my 750 and the Xeons (I think) it is 16x. I will try an X3450 in a minute, I'll see whether it's 16 or 18 then.
LGA1156 CPUs also don't need uncore to be >= 2x Dram, >=1,5x Dram is enough for them.

[123bob]

Not weird at all bob, if the board is really doubling all the mults - 185*20 is 3700Mhz, after all
What it should do, however, is let you change uncore from 12x to 40x or so. 12x being the slowest at 12x185=2220Mhz.

On the LGA1156 CPUs, the uncore mult is locked. For the 860 it's locked to 18x, for the smaller ones like my 750 and the Xeons (I think) it is 16x. I will try an X3450 in a minute, I'll see whether it's 16 or 18 then.
LGA1156 CPUs also don't need uncore to be >= 2x Dram, >=1,5x Dram is enough for them.

The bios is definitely busted. EDIT: WELL MAYBE CONFUSING AT LEAST. SEE BELOW :END EDIT Bclk is set to 185, for 3.7 gig on the CPU. I have uncore bios options of auto, then 10x to 24x. When I set 10x, it goes to 3.7gig (which is 20x185). It shows that in both the preview and in CPUz. When I set 18x, it goes to 6.66 gig (which is 36x185). I had set that before, and of course it couldn't even boot. Had to clear CMOS. So, there's an extra 2x uncore multiplier in the MSI platinum SLI board even using bios 3.6, which is the current one? The Auto setting seems to be stuck at a true 16x.



Obviously, I won't be acquiring another one soon....

Bob

EDIT: Did some more playing around. Found that if I take the mem multiplier from auto to x4, then I am given the option for a x9 uncore multiplier in the uncore multi section of bios.... So, got ram at 1480 and uncore (NB in CPUz) at 3330. Now it matches the 860 rig. I'll try to get 1T timing, then see if it's stable. So much for getting anything done on the day job. Still, it will be Gigabyte on the next rig..... :END EDIT

[123bob]

OK, now I should be straight.....

Here's the setup screenie for the 920 with the NB boosted and timing at 1T. This is loaded with BOINC.



New test run on this rig will begin soon with the exact same 24 WUs as before. I'm going to try not to abort the current WUs I have running on the 920 now, just move the folder out of the way and put the test folder in it's place. It should work....

Bob

EDIT: OK, it's crunching away as of 14:40 PDT....

[v0dka]

Nice, lets hope its stable.

[[XC] Hicks121]

Wow I didnt know you can run 1t CR on these.

Man I need to study.

jc & Bob...you guys rock with the detail & the info you provide.

[jcool]

Nice work bob.. and +1 on your next board NOT being an MSI

[DAK1640]

Very nice work indeed 123bob...Purchased my next CPU because of all you've done. Enuf already...

[123bob]

Nice work bob.. and +1 on your next board NOT being an MSI

Now that you mention it.......I stopped over at Fry's to look at back up drives. Wouldn't you know it, they have an 860 with a Gigabyte UD4 on sale this week for $379, a Thermaltake 750w supply for $99, and 4 gig Kingston 1800 sticks for $99. Smells like a cruncher to me....


@All, as to doing the work, I don't mind doing it. I was just as curious about it as everyone else. I think we can conclude testing at this data point when this run is over. What I want to see next is my max clock and it's power figures. This current point at 3.7 may be the sweet spot for power vs ppd on the 860, IDK.

One thing I'm really curious about now is how well the X3440 clocks.

The WUs are crunching away on the 920.....

Bob

[INFRNL]

I will try to update you on the X3440 this week! I am not the greatest at overclocking, but I might be able to give you a rough idea . I know it should easily do 3.8ghz with a good cooler (Megahalem on the way!!!!). will report back asap

Thanks again for your dedication BOB

$315 X3440/UD2 coming this week....

[123bob]

I will try to update you on the X3440 this week! I am not the greatest at overclocking, but I might be able to give you a rough idea . I know it should easily do 3.8ghz with a good cooler (Megahalem on the way!!!!). will report back asap

Thanks again for your dedication BOB

$315 X3440/UD2 coming this week....

...And I'm thinking you're going to have the most cost effective rig.....$315, dang that's not a whole lot....

Certainly the folks here can help with the clocking. I'm not the best at it either, but I'm getting the hang of i7 now.

Bob

EDIT: A further thought...For around the $315 plus the cost of two sticks of cheap DDR3, I might be able to cost justify upgrading the rigs in the kents farm. If I catch the timing of a sale right, it might be do-able. At least some of them... I'll have to look into some of the B3 kent's power draws and PPDs. It could be a no-brainer upgrade. :END EDIT

[INFRNL]

Keep in mind that $315 will be effective in about 50some days when I get my $18.xx from bing. Also this is just for mobo and cpu, but still damn cheap.

As for the day job comment, is that a day job, or like mine...an ALL day job

as for the uncore/nb freq. statement: how true is it that it makes a difference? I do not know if something has changed this weekend while screwing around with my rigs. obviously my ppd has fallen dramatically
but I just noticed that my nb freq on main rig is @3800 while one of my other rigs is @2880. I do not see much of a difference in production.

I thought the one that is @2880 was producing a little better, but I think for the most part its maintaining pretty good production

When I get time, I will have to play around a little bit to see if I can improve my ppd. I also noticed that my W3520 is runnin @4ghz@1.232v...I will have to check the bios as I thought it was running @ 1.3x v

Oh well thats another story for a different day

[fallwind]

Interesting results to say the least! We'll see if the higher uncore speed and 1T CR can make up that 0.3% difference. Only difference now is dual channel vs triple. We always knew it had little to no effect, but your final results should show just how little.

[Jaco]

What's wrong with MSI ?
I prefer Gigabyte myself , but never had issues with MSI boards .
Remember the MSI K8N Neo2 (socket939) ? that was a great board.

Usually MSI boards are pretty good performers imo.

[emuexport]

Well thanks to Bobs dedication and stickling for procedures I was able to duplicate it perfectly! (unlike putting my machines on another account! )

Anyways both my i7 860 and 920 now have the same WUs and are now crunching these . I made a few changes to Bobs procedure though see bold below (I never could follow the rules! )

OK, I did it.

Here's how.

The 860 rig is the "host" machine.
The 920 is the "client" or "target" machine.

1. I suspended crunching on both. (activity - Suspend and Network Activity Suspended)

2. Aborted the current running WUs (Yes, I tried REAL hard not to scream... (This really pained me too! But better aborting them than losing them completely least this way the server will give it out to someone else!)

3. I aborted all "waiting to run" WUs on the 860 rig, except for 24 HFCC WUs. These are the test WUs.

4. "Updated project" so the queue was cleared. Only 24 units sitting there.

5. I aborted ALL the WUs on the 920 rig. (With the same screaming as above. I think the rig screamed louder though....) (This is where I changed tactic and simply put BOINC in activity - Suspend and Network Activity Suspended. Then exited/closed Boinc and made a copy of the Original files and then deleted the contents of the BOINC folder, pasted the 24WUs BOINC folder from the host and presto on starting BOINC it had the same WUs)

6. "Updated project" so the entire queue was cleared. No WUs in this rig. (No need for this step if you follow my suggestion above)

7. I copied the entire "BOINC" folder, under the app data folder on the 860 rig, on to a USB stick.

8. I made a backup of the BOINC folder of the original 920 rig so I can recover it later, I hope.....

9. I copied the BOINC folder from the 860 rig into the app data folder of the 920 rig.

10. I disconnected the network cable from both rigs. (Suspend network activity)

11. Rebooted the 860 rig and resumed crunching. It started on the test units. (No reboot just run BOINC again)

12. Rebooted the 920 rig, it then showed the test WUs. Ran benches, which turned out the same as reported before, then resumed. It is now crunching test WUs.

[123bob]

OK, results first and then I'll post to some of the other questions/comments above.


The only difference now is the 920 has triple channel and the 860 has dual channel RAM. Screenie of the new 920 setup posted above, again.





Here's the raw data of the 920 matched exactly to the 860. (Except, of course, for triple vs dual channel RAM.)



As before, you are welcome to check my math here....

Doing the same math as above we get 84:639:806.

Converting to hours that's 94.873888. Dividing that by 24 WUs we get 3.9530787 Hours/WU.

That's .042488396 hours FASTER than the 860. Converted to minutes, the 920 is now 2.54930 minutes per WU FASTER than the 860.

Perhaps the triple channel does come into play?

Perhaps there is some natural variance in the numbers? I'll make one more 920 run at these settings to see. We don't know how consistent the numbers are without multiple tries......

If there's not much variance between runs, we can make a couple of conclusions. (OR at least I can. You may, or may not, agree....)

1. Even at 2.5 minutes per WU difference between rigs, the 860 is still in the same league with the 920. The difference is just over 1%. The back end variances of WU score, quorum wingmen, quorum score, and WU daily timing are MUCH greater than that..... 1% difference would be lost in the noise. The 860 is more power efficient. For those with high power rates, this is a no-brainer.

2. QPI setting makes a pretty good difference in production. This was shown when I had the QPI cranked down on the 860 and the rig was fairly well behind using the older "random WU average" method before. I could test this using the new "fixed WU" method, but I'm convinced it's real.

3. Uncore/NB and/or the 1T timing change makes a reasonable difference. The 920 with the faster Uncore and 1T timing was 3.21 minutes/WU faster than the lower uncore and 2T timing. This is about 1.3% faster. Personally, I would bet it's the uncore that made more of the difference, over the 1T vs 2T timing change.

4. The new "fixed WU" method of testing rocks!! Thx to Snowcrash for suggesting it and everyone else for fine tuning it. I'll be keeping those test WUs on the stick for a long time....I may have to run them on some of the kent's farm rigs when they come home. What I'll have to keep in mind now is the difference between threads and cores. The i7 rigs do 8 threads at a time vs the kent's 4 cores.

Regards,
Bob

[jcool]

First of all bob, let me thank you again for your time and dedication put into this, truly amazing

Below is just my 2 cents on your conclusions.

1. Even at 2.5 minutes per WU difference between rigs, the 860 is still in the same league with the 920. The difference is just over 1%. The back end variances of WU score, quorum wingmen, quorum score, and WU daily timing are MUCH greater than that..... 1% difference would be lost in the noise. The 860 is more power efficient. For those with high power rates, this is a no-brainer.

I agree, they are equal crunchers, as was expected. The 860 may be more power-efficient at 3,7Ghz, however, judging from the LGA1156 and 1366 CPUs I have tried until now, this picture changes dramatically if you go to around 4Ghz. Almost all Lynnfield CPUs seem to require vast amounts of Vcore to handle 4Ghz, 1,4V seems like average, while most D0 920's can do 4Ghz at around 1,25V. I compared an X3450 running 4Ghz (bad one.. 1,43V and crunching at 87C for this test ) to an average i7 and came up with ~44W less for the 920 with otherwise identical setups (PSU, GPU, SSD identical).

2. QPI setting makes a pretty good difference in production. This was shown when I had the QPI cranked down on the 860 and the rig was fairly well behind using the older "random WU average" method before. I could test this using the new "fixed WU" method, but I'm convinced it's real.

Yes, using a high QPI mult seems to be crucial for Lynnfield performance. However, I could not detect the same behaviour with Bloomfields, I ran my 975 at different BCLKs and QPI mults over the last half year or so, and I never noticed any changes in PPD or WU runtimes caused by different QPI bus speeds. Of course I didn't do a real scientific test, but I still don't think Bloomfield cares about high QPI mults the same way Lynnfield does. maybe someone wants to try it out?

3. Uncore/NB and/or the 1T timing change makes a reasonable difference. The 920 with the faster Uncore and 1T timing was 3.21 minutes/WU faster than the lower uncore and 2T timing. This is about 1.3% faster. Personally, I would bet it's the uncore that made more of the difference, over the 1T vs 2T timing change.

Yeah, most probably it is the higher uncore speed giving the boost, as 1T vs. 2T does next to nothing on Intel X58 chipsets, not even if you run Everest mem benches etc.

However, I must revise my earlier statement (that Uncore speed = L3 speed). It appears that the L3 is synched to the CPU's core speed, ie Core speed = L3 cache speed. I apologize for the misinformation, I read that on some tech site a while ago, obviously they were wrong. The Intel spec finder lists L3 speed = core speed for all Core i7 models, so I guess that proves it.

[123bob]

1. As for the day job comment, is that a day job, or like mine...an ALL day job

2. as for the uncore/nb freq. statement: how true is it that it makes a difference? I do not know if something has changed this weekend while screwing around with my rigs. obviously my ppd has fallen dramatically
but I just noticed that my nb freq on main rig is @3800 while one of my other rigs is @2880. I do not see much of a difference in production.

I thought the one that is @2880 was producing a little better, but I think for the most part its maintaining pretty good production

1. Sometimes it's a day and night job. Speaking of that, I have some serious catching up to do....

2. The results are in the above post. It seems to me that QPI makes a pretty good difference. The Uncore/NB difference is there, but not as pronounced.

I figure if we know this, we all can tweak for it and get the most out of our rigs. I would suggest none of us go "ape crazy" to push these, just be aware that they should not be ignored.

What's wrong with MSI ?
I prefer Gigabyte myself , but never had issues with MSI boards .
Remember the MSI K8N Neo2 (socket939) ? that was a great board.

Usually MSI boards are pretty good performers imo.

I have had very good luck with the MSI P45 Neos I still run on the kents farm. Those work fine.

However, with these i7 boards, for some reason I find them confusing in the mechanics of the bios. For instance, to change CPU vcore, you have to push the +/- buttons on the numpad, not next to the backspace key. No where in bios or the book do they tell you this. Page up/Page down also works. I have a tiny laptop keyboard on the racks and that took me a bit to figure out. I had to ask the question at the WSI forum to figure it out. Talk about feeling like a noob...

Another example. The X58 Platinum SLI comes with OC switches on the board. It overwrites your bios on boot. The intention is for auto overclock. It comes set for 133 bclk. No where in the manual does it say how to turn the damn thing off so you can bios OC. I found that by trial and error. Very irritating.

The "one button OC" feature on the 860's P55-CD53 mobo didn't work worth a dang. It just kept rebooting when I tried. Bios OC shows that the system is very capable of OCing, just that the board couldn't come up with any combinations it liked on its' own. It's not a feature I really care about, but I feel sorry for the noobs that would like it to work....They paid for the feature.

I think they just have some kinks to work out. Right now, from what I'm seeing from others, for the same money, the Gigabytes seem better.

Remember, this is just my opinion, based on ownership, right now.

Well thanks to Bobs dedication and stickling for procedures I was able to duplicate it perfectly! (unlike putting my machines on another account! )

Anyways both my i7 860 and 920 now have the same WUs and are now crunching these . I made a few changes to Bobs procedure though see bold below (I never could follow the rules! )

Your tweaks look good. I realized after I blew all the WUs out of the 920 (or "target" rig) that I could have just moved them out for a day and put them back. That's a good improvement to the original write up.

I disconnected the net cable just to be REAL sure my test WUs didn't escape on me... I had the horrible picture of them escaping from the 920 rig, getting out in the wild and possibly bringing the server down, or destroying the planet..... I've since had the chance to see how "Net Act Supend" works and will likely do that in the future.

Regards,
Bob

[123bob]

I agree, they are equal crunchers, as was expected. The 860 may be more power-efficient at 3,7Ghz, however, judging from the LGA1156 and 1366 CPUs I have tried until now, this picture changes dramatically if you go to around 4Ghz. Almost all Lynnfield CPUs seem to require vast amounts of Vcore to handle 4Ghz, 1,4V seems like average, while most D0 920's can do 4Ghz at around 1,25V. I compared an X3450 running 4Ghz (bad one.. 1,43V and crunching at 87C for this test ) to an average i7 and came up with ~44W less for the 920 with otherwise identical setups (PSU, GPU, SSD identical).

Agreed. Something less than 4 gig, so far, seems to be the 860 sweet spot, based on my data and Emu's 4 gig power data. The key will be in the production numbers. If 3.7 vs 4 gig makes substantial production differences, then the scale tips to the 920 for some people's cases.

I'm going there next.

See ya'll at 4 gig....
Bob

[v0dka]

If 3.7 vs 4 gig makes substantial production differences, then the scale tips to the 920 for some people's cases.

That's an 8% clock difference, and since we know that performance per clock is the same also 8% greater production. For just a higher investment in hardware I'd say that is a nice bonus.

But then you have the power consumption. Personally I would always go for the 860 at a reasonable 3.7Ghz and save ~40w. Crunching is all about good performance for the lowest possible costs.

Also, it's not only the inital investment in the 860 system that is lower compared to the 920 system, I think you can increase that difference when you realise that a budget board and budget RAM are good enough because you are not going to push it past 3.7. The 920 pretty much needs high end stuff.

[123bob]

I agree with you V0dka, for pure cruncher duty. In fact, I think INFRNL's choice may be a winner for that. I run my pure farm rigs so they don't need a babysitter. I want them to run, run, and just run. If I can have that and power efficiency, I'm interested.

However, for some folks using rigs for games, applications, and such, the 920, the 860, and budget come into play. I can see that aspect too.


In other news......Well, I tried my best to get 4 gig out of the 860. I couldn't even get 3.885gig stable on it. (185x21)

I tried the max vcore the bios would let me set. I mean I had CPUz showing 1.536v at idle! I could boot, but no primes....

I jacked up all the other voltages to what they had for max safe, dropped the QPI to x16, still no love. This particular chip ain't going there.....

I dropped it all back to the magic 3.7 point and it's fine. No harm done from the voltage, at least right now....

So, it's up to Emu to give us the 4 gig point. Best of luck to you my friend!

Regards,
Bob

[don_xvi]

When figuring the 860 vs 920 equation... there's also the hexacore factor, yes ?

I was really excited to go 860 until I figured out that the upgrade path for 1156 seems to be dual core, whereas the way to get hexacore is to go 1366. That made the decision for me, I'm going with a 920. Then I have to hope there's a reasonable hexacore some day soon enough to make it worthwhile, otherwise I'll wish I'd gone 860. There's always that waiting option... NAH !!!