Seller Listing of D0 Stepping Core i7 920's

[ben805]

I think Ben085 has one ....

Yup, mine is the batch 3849A846 I think it's a decent clocker, on air it can do 4.50Ghz stable in SuperPI 32M, at 4.0Ghz with HT enable it only needed 1.22vcore to pass LinX, Prime, and 3D tests. On water, I run 4.40Ghz HT enable can pass LinX with 1.368vcore but to pass Prime it needed 1.40v. At the moment I'm running LinX at 4.50Ghz with HT off, the heatload is unsustainable with HT enable at 4.50Ghz. will have do some benchmarks to compare 4.40 (HT on) vs. 4.50 (HT off) to see which one I would chose to run 24/7. I have 12GB of ram too, other than running about 2'C hotter than 6GB, it doesn't seems to affect my overclock, I'm loving this D0...they are off the chain man.

[Golly]

Heres mine on air, I was trying a few things and didn't notice that I had switched to "SPD" instead of XMP for mem timings, when I get a chance I'll redo it

http://valid.canardpc.com/show_oc.php?id=562903

That's got to be number one in the world!

[theonlybabyface]

They said earliest D0s OC the BEST! You should try it.

Well duh! What forum do you think this is? Just wanted to see if anyone has this stepping yet so I know what to expect. My C0 does 4.2 but at 1.4v...

[ColonelCain]

Well duh! What forum do you think this is? Just wanted to see if anyone has this stepping yet so I know what to expect. My C0 does 4.2 but at 1.4v...

He was talking about how the D0 stepping i7's that were produced first seem to OC better than the ones that were manufactured recently...

[hlonipha]

Yup, mine is the batch 3849A846 I think it's a decent clocker, on air it can do 4.50Ghz stable in SuperPI 32M, at 4.0Ghz with HT enable it only needed 1.22vcore to pass LinX, Prime, and 3D tests. On water, I run 4.40Ghz HT enable can pass LinX with 1.368vcore but to pass Prime it needed 1.40v. At the moment I'm running LinX at 4.50Ghz with HT off, the heatload is unsustainable with HT enable at 4.50Ghz. will have do some benchmarks to compare 4.40 (HT on) vs. 4.50 (HT off) to see which one I would chose to run 24/7. I have 12GB of ram too, other than running about 2'C hotter than 6GB, it doesn't seems to affect my overclock, I'm loving this D0...they are off the chain man.

Talk about loving this DO's. I'm stable now @ 4.3ghz ht on with 1.34v. I will do some more tweaking tomorrow. will try 4.4 ht on.

[Kougar]

Hlonipha, could you please try something for me since we have an identical batch 920? I'm curious why I need so much more voltage than you...

Hours of Prime95 never showed any problems, also LinX running at the default problem size of 768 passed 25 runs error free. But, if I tried running LinX, and set it for "Problem size 25083" (this requires you have 6GB of RAM btw), and let it run for 10 runs... the system would eventually blue screen ~10 minutes in.

If you can pass 10 (or even 20 runs, but this requires an hour and 20 minutes at this problem size) then I'd love to see how you configured the BIOS! Your temps are about 8c better than what I'm getting at 4.2GHz just now.

[Garrett]

Yup, mine is the batch 3849A846 I think it's a decent clocker, on air it can do 4.50Ghz stable in SuperPI 32M, at 4.0Ghz with HT enable it only needed 1.22vcore to pass LinX, Prime, and 3D tests. On water, I run 4.40Ghz HT enable can pass LinX with 1.368vcore but to pass Prime it needed 1.40v. At the moment I'm running LinX at 4.50Ghz with HT off, the heatload is unsustainable with HT enable at 4.50Ghz. will have do some benchmarks to compare 4.40 (HT on) vs. 4.50 (HT off) to see which one I would chose to run 24/7. I have 12GB of ram too, other than running about 2'C hotter than 6GB, it doesn't seems to affect my overclock, I'm loving this D0...they are off the chain man.

Reading this has me all fired up man
Anyone please give me advice 3847A or 3849A... I can't choose lol... seems like a luxury problem to me

[Zucker2k]

Reading this has me all fired up man
Anyone please give me advice 3847A or 3849A... I can't choose lol... seems like a luxury problem to me

Dude, find the money and buy both. Keep the best and sell the other after testing. Since it's a D0 you'll probably break even or even make a profit. I made a profit on my open box and got a xeon with change.

[The Water Dog]

i got the 3847A and are on 3,6 with everything on stock. I have sett all volts on my own, but they are on "normal".

[roller11]

Im buying Core i7 in Amazon purche date may 4

SLBEJ

Batch 3849A925

Are you saying that you bought and received a D0 batch 3849A925 from Amazon.com in the USA? Or did you mean that you have reason to believe you will get a D0 from Amazon.com, but you haven't received it yet?

[Renegade5399]

Here's the results of mine so far. The Xeon is an excellent chip. No options in the BIOS are locked out and I have all the same QPI and RAM dividers I had with the C0 920.

My results thus far.

Overall I am very happy with this purchase and sice these can be had for about the same price as the 920 (sometimes cheaper) I say if you can get one, go for it.

[tekjunkie]

Here's the results of mine so far. The Xeon is an excellent chip. No options in the BIOS are locked out and I have all the same QPI and RAM dividers I had with the C0 920.

My results thus far.

Overall I am very happy with this purchase and sice these can be had for about the same price as the 920 (sometimes cheaper) I say if you can get one, go for it.

This is defintely some good news(doing uncore over 4GHZ stable), hopefully the Classified or BR(the board that I have) would follow suit with their BIOS updates.I should have gone with the DFI over the BR for my second board.I went for some bling.MEH!!Hopefully the BIOS issues will be ironed out.

EDIT: Do you have to mess with the RTL values in the BIOS to get 2:10 @ that low VTT(1.3)??

[Renegade5399]

EDIT: Do you have to mess with the RTL values in the BIOS to get 2:10 @ that low VTT(1.3)??

Yes I do.

Channel 1 Auto = 55

Channel 2 +1 = 56

Channel 3 +2 = 57

tREFI x4

Low VTT seems to be common for these D0's anyways. Another reason I wanted one.

[hlonipha]

Hlonipha, could you please try something for me since we have an identical batch 920? I'm curious why I need so much more voltage than you...

Hours of Prime95 never showed any problems, also LinX running at the default problem size of 768 passed 25 runs error free. But, if I tried running LinX, and set it for "Problem size 25083" (this requires you have 6GB of RAM btw), and let it run for 10 runs... the system would eventually blue screen ~10 minutes in.

If you can pass 10 (or even 20 runs, but this requires an hour and 20 minutes at this problem size) then I'd love to see how you configured the BIOS! Your temps are about 8c better than what I'm getting at 4.2GHz just now.

Sorry for the delay, I was out all day in sunny Miami after a crazy night!! Will try this and report back. Any reason why that specific size of a problem?

[Kougar]

Here's the results of mine so far. The Xeon is an excellent chip. No options in the BIOS are locked out and I have all the same QPI and RAM dividers I had with the C0 920.

My results thus far.

Overall I am very happy with this purchase and sice these can be had for about the same price as the 920 (sometimes cheaper) I say if you can get one, go for it.

Holy crap that's impressive... good temps as well, but I just can't believe your vcore setting! I see you DO have HT off, but still... I've played around with everything now but nothing appears to let me keep the system stable at 4.2GHz except vCore. On the plus side I can keep VTT and IOH voltages much lower than Anandtech's D0 chips. Did you modify PLL, clock skew, or clock drive settings any? And what program gives you all the information in the right-side bar just to check? Thanks for running that specific test, I didn't think it was possible on that low a vcore.

Sorry for the delay, I was out all day in sunny Miami after a crazy night!! Will try this and report back. Any reason why that specific size of a problem?

I don't know specifically, but it was the only LinX test my system would fail (BSoD) at. I presume because it places the largest load upon the CPU/RAM, and the larger the problem size the longer the duration of the load upon the CPU. Just like IntelBurn, CPU load alternates between 100% and 0% loads but the duration of the load increases as the problem size goes up.


Screenshot: IMG Link (1920x1200)


I am still in the process of lowering my VTT & IOH voltages to find the lower limits, so far I'm at 1.42 vCore, 1.20v IOH, 1.30v QPI/VTT, 1.66v RAM. Several hours each of LinX, Prime95, and Left 4 Dead are stable at the above settings. Also used Memtest to verify the RAM. Hyperthreading is ENABLED. Uncore of 3800MHz is extremely unstable, I can't even run it at Renegade's frequency! Once I bottom out the voltages I will try an Uncore of 3600MHz and see how it goes on minimal voltages, but I expect 3400 will by my 24/7 setting...

Even with a QPI/VTT of 1.38 and IOH of 1.30, and uncore of 3200, if I lowered the Vcore to 1.40v the system would BSOD ten minutes into 5GB problem sizes in LinX. I'm not sure what else is left to try, the only thing I have not attempted were the PLL settings or maxing out the CPU Clock Drive setting. I am very happy to have a fully stable 4200MHz OC with these RAM timings, unusually high vCore or not.

[Renegade5399]

Holy crap that's impressive... good temps as well, but I just can't believe your vcore setting! I see you DO have HT off, but still... I've played around with everything now but nothing appears to let me keep the system stable at 4.2GHz except vCore. On the plus side I can keep VTT and IOH voltages much lower than Anandtech's D0 chips. Did you modify PLL, clock skew, or clock drive settings any? And what program gives you all the information in the right-side bar just to check? Thanks for running that specific test, I didn't think it was possible on that low a vcore.

With my DFI board and the 4/28 BIOS, the mem bandwidth actually decreases with HT on, plus I haven't ever really seen a real need for it. The extra 8-10°C in temps is not worth the supposed benefit.

My VTT and IOH voltages are just rough settings. I can tune them later, but for now I'll just leave them. To determine the VTT needed, I set the desired BCLK in the BIOS, then boot to memtest. If memtest locks right up or doesn't complete test #2 properly (test runs past 100%), then I need more VTT. For 2000MHz RAM, I am very happy with the low VTT I need. My C0 920 needed 1.58v VTT to hold 2GHz RAM speed stable. My IOH voltage is what I have found to keep me 3D bench stable with these dual GTX280's. Don't know why, it just does.

PLL is left at default 1.8v. PLL is a non-factor when compared to C2D chips. It really doesn't affect much at all in my experience.

Clock skew was not touched at all either.

Under the RAM timings, using CPU Tweaker, I set my subtimings accordingly. This latest BIOS allows me to set the tRTL of each of the 3 channels. This has made a huge difference is RAM stability at high speed. The BIOS also allows me to set the tREFI higher to add to the stability.

The program that displays the info is called Everest. That's the OSD function of the program.

I don't know specifically, but it was the only LinX test my system would fail (BSoD) at. I presume because it places the largest load upon the CPU/RAM, and the larger the problem size the longer the duration of the load upon the CPU. Just like IntelBurn, CPU load alternates between 100% and 0% loads but the duration of the load increases as the problem size goes up.


Screenshot: IMG Link (1920x1200)


I am still in the process of lowering my VTT & IOH voltages to find the lower limits, so far I'm at 1.42 vCore, 1.20v IOH, 1.30v QPI/VTT, 1.66v RAM. Several hours each of LinX, Prime95, and Left 4 Dead are stable at the above settings. Also used Memtest to verify the RAM. Hyperthreading is ENABLED. Uncore of 3800MHz is extremely unstable, I can't even run it at Renegade's frequency! Once I bottom out the voltages I will try an Uncore of 3600MHz and see how it goes on minimal voltages, but I expect 3400 will by my 24/7 setting...

Even with a QPI/VTT of 1.38 and IOH of 1.30, and uncore of 3200, if I lowered the Vcore to 1.40v the system would BSOD ten minutes into 5GB problem sizes in LinX. I'm not sure what else is left to try, the only thing I have not attempted were the PLL settings or maxing out the CPU Clock Drive setting. I am very happy to have a fully stable 4200MHz OC with these RAM timings, unusually high vCore or not.

I wish I could be of more assistance, but I'm not familiar with the Gigabyte boards at all.

[hlonipha]

Here you go m8. It ran with a smaller problem size than you asked due to my memory usage on vista64. Linx would not allow me to select 25k+. I think I got a nice chip here. Have not even tried to tweak with HT off. I think I will start today to see my ht off max clocks. vcore idle 1.36v load 1.34v. it takes even less v on bios f5g. not sure why but f7d does not have the dual cold boot problem so I will keep it for now.

[ben805]

Nice. usually mine takes 50min to run only 5 passes of LinX with 12GB of ram, problem size around 36000 and used up almost 11GB of ram. After you're done with LinX make sure you hit up Prime95 as well, if your chip behave anything like mine you will most likely need to bump up the vcore some more to pass couple hours of Prime95

[tekjunkie]

Yes I do.

Channel 1 Auto = 55

Channel 2 +1 = 56

Channel 3 +2 = 57

tREFI x4

Low VTT seems to be common for these D0's anyways. Another reason I wanted one.

Thanks. I have been saying this(that the RTL is the key for high memory clocks) and I hope some of the other boards(Foxconn hint hint) will follow suit with having adjustable RTL values in the BIOS.

As far as the VTT goes for D0, yes it does require low VTT, but only if you are doing 2:6 or 2:8. Once again it comes down to the board support with the RTL adjustment to run high memory clocks.Fo instance on my BR with my 920 DO, I could do 200x21 2:8 real easy with both LinX and prime stablity, but no matter how hard I try can't do 2:10 prime stable.It is Linx Stable allright(30 passes), but an overnight prime run crashes with a 124 BSOD.

[anzial]

You guys are taking this thread completely off topic. You can discuss the overclocking issues elsewhere, lets restict this thread to locating these awesome cpus, mkay?

[hlonipha]

Nice. usually mine takes 50min to run only 5 passes of LinX with 12GB of ram, problem size around 36000 and used up almost 11GB of ram. After you're done with LinX make sure you hit up Prime95 as well, if your chip behave anything like mine you will most likely need to bump up the vcore some more to pass couple hours of Prime95

Prime was done already my friend . look at a couple of post above this one. I ussually do prime before linx. Same volts

[Kougar]

Renegade, Hlonipha, thanks for the posts guys! Admittedly memory subtimings are beyond me for the moment, I'll have to read up on those later. For 1600MHz to be stable at 7-7-7-18 1T I have not had to modify any of them thankfully.

I have another question... I have noticed vdroop is incredibly high on my motherboard. Are both of you enabling LoadLine Calibration or something else? Usually I thought CPUZ would show vCore below what was set in the BIOS because of vdroop, but for both of y'all it appears to align pretty close or right at your BIOS vcore setting. Have either of you modified CPU Clock drive any? So far I've left it at the default 800mv setting. And what PCIe frequency are both of y'all using out of curiosity??

Anyone else getting problems with CPU-Z randomly failing to start? Something is blocking the driver from loading properly, it even affects CrystalCPUID from working. Ran another 10 hour test of Prime95, here's CPU-Z refusing to load and what's left of CyrstalMark: IMG Link

[tekjunkie]

It looks like ncix have dropped the prices for the D0 to $277.54 (lol the 920 without the D0 gurantee costs more) and in stock. Get them while they are available.

[hlonipha]

Renegade, Hlonipha, thanks for the posts guys! Admittedly memory subtimings are beyond me for the moment, I'll have to read up on those later. For 1600MHz to be stable at 7-7-7-18 1T I have not had to modify any of them thankfully.

I have another question... I have noticed vdroop is incredibly high on my motherboard. Are both of you enabling LoadLine Calibration or something else? Usually I thought CPUZ would show vCore below what was set in the BIOS because of vdroop, but for both of y'all it appears to align pretty close or right at your BIOS vcore setting. Have either of you modified CPU Clock drive any? So far I've left it at the default 800mv setting. And what PCIe frequency are both of y'all using out of curiosity??

Anyone else getting problems with CPU-Z randomly failing to start? Something is blocking the driver from loading properly, it even affects CrystalCPUID from working. Ran another 10 hour test of Prime95, here's CPU-Z refusing to load and what's left of CyrstalMark: IMG Link

Let's take this to the overclocking DO thread. have not touch pci-e or default cpu clock drive. it is set to auto & LLC is enable. Not sure about your cpuid problem. wait a few seconds after wondows loads completely before you start cpuid.

[Kougar]

Yeah, that's probably a good idea. I'll move the overclocking discussion elsewhere.