*** Official Gigabyte 8PENXP and 8IPE1000/MK/Pro (865PE) Thread ***

[Night Blade]

If anyone else has a Pentium4-A CPU with this board let me know please, would like to know if you can overclock at all. Only have a week or so to return this thing & I really don't want to, it's been 100% stable since installing it, that's more than I can say about the P4P800, my opinion of Asus changed after using that thing...

 

[computer]

Am I to understand the 8PENXP doesn't support ECC memory? Where are you guys buying the 8PENXP & how much? I can't find it anywhere at pricewatch. Thank you.

 

[computer]

^

 

[computer]

No one knows if the 8PENXP supports ECC memory?

 

[Auric]

New BIOS's have been released... not for the Pro though.

 

[brainwave]

Just wondering what brand memory everyone is using. I just ordered Corsair 512 TwinX 3200LL (2x256) to take advantage of the dual channel. I have been in SC so far. Some say Kingston is the best memory for Gigabyte boards but I got burned with a bad stick of Kingston in the past, so this time I went with Corsair, but I have not yet installed it. Mobo is 8IPE1000 Pro. Anyone using Corsair? Any problems? What timings?

 

[computer]

Anandtech's 865/875 mobo round-up shows Corsair as the BEST memory to use. Sorry I don't have the link now. I THINK it may NOT have been the 865/875 round-up, but a test of memory....I can't recall. But suffice to say that in every test, Corsair was the fastest followed by Crucial. But what I can't understand is they were using CAS2 Crucial memory which they don't even sell! I even asked them and they said they don't have any CAS2 DDR400. It could be the testers set the timing to those CAS2 specs. Hey I found the link, you can browse the article here, that page just happens to be on an Abit board. I see they were using the CAS3 Crucial after all and were setting it faster. My last check of 512mb DDR400 @crucial showed it to be $95 if I recall. MUCH cheaper than Corsair, so it's a better deal. I think the Corsair when I checked was $135.

 

[Auric]

As always, you cannot just choose a product based on a brand name because ultimately all that stands for is a guarantee and support. There are a limited number of companies that actually design and manufacture RAM chips -and even less when it came to DDR400 recently as Micron was not producing any and so its Crucial division used Samsung instead. Winbond has been the primary source for higher spec chips and that is what Corsair, TwinMOS, Kingston, Buffalo, et alli have used in most cases. Any of these are probably dependable as far as board quality and assembly so it comes down to price and whether you are willing to pay a premium for marketing and dubious extras like heatsinks, "matched pairs" and overclock tested parts (which all sold as faster than DDR400 have been).

 

[brainwave]

This review of the 8IPE100 mobo says it has options to enable "PAT-like performance" in the BIOS:

http://www.hardwarezone.com/articles/articles.hwz?cid=6&aid=808

"Since the GA-8IPE1000 has memory options that could enable PAT-like performance, we tested the board twice, once with the options disabled, and the other with the options turned on. These options were disabled in the BIOS by default, which is why we had to benchmark it twice in order to show the difference between the default and optimized modes. When these options were enabled, CTIA 2.7 (a software for checking the PAT status) reported that PAT was only enabled partially - which means that the GA-8IPE1000 won't be giving you 875P type of speed and performance."

I don't see any such BIOS options other than Advanced Chipset memory timing options that every board has.

What are they talking about????

 

[computer]

Yeah I know memory modules have individual chips of different brands on them, I was just relaying what the article said. At worse, it's a good starting point for getting good memory and it's safe to say that Corsair memory is probably going to perform the best....or else they would have never done the memory review. In the case of Corsair; regardless of chips used (if they even use different chips on their "Corsair LL TwinX (CAS 2)") I believe it would be safe to say that all of these specific Corsair modules with that nomenclature have to meet certain established parameters regardless of the brand of different chips used (if any).

 

[brainwave]

There is a new bios for the 8IPE1000 Pro, F7d. This BIOS corrects the memory timing, adds the option for CL 3.0, and allows the CL 2.0 option to override the SPD setting (which F6 bios did not allow). So far I have had no problems with this bios and my SANDRA memory bandwith jumped 100 points.

 

[Auric]

Where did you get the info about it? What is the memory timing issue? How did you load it? I used QFlash and it appears to load successfully but does not boot and upon reset loads the backup BIOS. I guess I will have to wait for a non-beta. Although I can't say I had any problems before I am just a glutton for upgrade punishment.

 

[brainwave]

Hi Auric,
BIOS F6 allowed CAS latency (CL) options of 2.0 or 2.5. However at 1:1 ratio, using DDR400, CPU-Z showed that CL remained 2.5, even when I set the CL to 2.0. However, at 4:5 ratio, CL did read 2.0, but I was only using that ratio when oc'd. I wanted my Corsair Twinx3200C2 at its advertised timings of 2-3-3-6 even when not oc'ing.

Since upgrading to F7d BIOS, I am now able to get CL of 2.0 with 1:1 ratio. Also, there is a new option of CL 3.0 with this new bios so it's obvious Gigabyte has modified and improved the mem timings.

I loading F7d using Qflash from within BIOS, which does not require using floppy or Flash file. After the first time I did it, the computer rebooted and then reloaded F6 from the backup bios (thank you Gigabyte for dual bios, which has saved me several times!). I discovered that my memory timings were set to 2-3-2-6 rather than SPD. When memory was set to SPD (2.5-3-3-8) the new bios loaded no problem. Then I was able to reset timings to 2-3-3-6 without problem. I think the timings need to be relaxed whenever you mess with the bios.

With the new bios F7d, Corsair TwinX3200C2 at 2-3-3-6 1T, my SANDRA mem bw went from about 4325 to 4470 (buffered scores). I have not yet oc'd with the new bios. I was able to get my P4 2.4c up to 2.9 no problem with F6 bios. I want to be sure everything is stable for a while before I try oc'ing the new bios. Plus, I was hoping to hear other people's experience with this bios, but this thread has been awfully quiet. Considering that this is a great, stable 865 mobo I hope more people will post their experiences.

 

[Ferrari355]

Are 8IPE1000 Pros good overclockers? I bought a 2.4c from someone who was able to run it at 3.4. Can the 8IPE1000 oc that high?

 

[Auric]

Thanks for the info brainwave. I will have to give it another go but try restoring default BIOS settings first. My 2.4C is also running at 2.9 with the 1.6 divider leaving the memory at spec. Did you notice in CPUID that the F6 BIOS is listed as F4?

 

[brainwave]

I was able to oc my 8IPE1000Pro with 2.4c up to 3.2 (F6 bios), but with my 3200C2 DDR, it was not as stable as I would like. (I have zero tolerance for anything less than "rock stable.") That is, I could boot into WinXR ok, but some programs crashed. I think with faster memory, like 3500 or 3700 you could prob oc this board to 3.4. On the other hand, if you were looking to buy a board in order to oc a 2.4c to 3.4, I would prob buy an ASUS. I bought this Gigabyte board because it is very stable, it has more features (like firewire) than any other board at this price, it does oc very well, and it has dual bios, which has been very helpful several times. I had an Albatron mobo without dual bios (some Albatron's do have dual bios) which corrupted when flashed and I had to RMA the whole board. That was a pain.

About CPUID: Both CPUID and CPU-Z read the F6 and F7d bios as F4, even though it reads the bios date correctly. I read some other threads about other mobos in which people were saying CPUZ was misreading the bios version. Maybe when you flash the bios it doesn't change the identifying label. I trust the bios version # which shows during boot.

I am very happy with this board. I think the reviews of it are good, but not overly enthusiastic because Gigabyte did not activate PAT on this i865 board. As a result, others like the Abit and Asus, out-perform it on benchmarks (but not by much). But in everyday use, a 3-5% difference in performance is not going to be noticeable. On the other hand I have read many unhappy comments on threads and Newegg reviews about the Abit IS-7, especially the sound quality and stability. A fast board is useless if it is unstable. The 8IPE1000 is plenty fast, and I would gladly give up 10% of performance for stability.

 

[Auric]

I feel the same way re: stability. I also like the layout of this board as said before with plenty of room for heatsinks, and with proper mounting of the NB HS. After reading of the audio problems and heatsinks falling off the Abit I am doubly glad I did not get it. It would be nice if Gigabyte did offer PAT. Do you know what the Performance BIOS option does and those three settings at the bottom of the Advanced Chipset page for the memory (ack I should have wrote them down but maybe you know what I am referring to -I don't think they are in the manual)?

The BIOS flash went smoothly after restoring default. About the version, I meant to say CPU-Z as that is what I use (I used CPUID in the past), but same result it seems. I know to trust the version shown at boot but just thought it odd that it was being read wrong.

 

[brainwave]

I understand that Top Performance (TP) "enabled" simply overclocks the CPU slightly. I have read about stability problems if Top Performance is enabled when you manually oc the board, so I prefer to keep TP disabled and oc the board slightly using the frequency controls manually.

The 3 settings at bottom of Advanced Chipset screen: these are not documented anywhere by Gigabyte, but in my various reading I have figured out that the 1st one (Command per something) sets mem timing either 1T (enabled) or 2T (disabled). The default is disabled, but it should be set to enabled. The 2nd one (Fast chip something) defaults to disabled. After I changed it to enabled, CTIAW (which is a program that tells you if PAT is enabled) changed from "PAT disabled" to "PAT partially enabled," which sounds pretty good to me. The 3rd setting (I forget what it is labelled) is enabled by default and I do not know what it does.

I did read one review which says this mobo is capable of "partial PAT." I think it was on HardwareZone.com. If I can find it again I will post the link later, but I am busy today trying to fix my neigbor's messed up old Gateway machine...wish me luck-- lol!

 

[tallman45]

I am looking into getting this mb and read on the Asus website that the Sata Raid only works with Windows XP. I have Win2K pro and want to save on buying a new OS. This may be specific to Asus but want to make sure

1. Is this a requirement of the ICH5R or is it specific to this MB ?
2. Can I run a single Raptor hd for OS/Apps?

Thanks

 

[Auric]

The settings are:

Command Per Clock
Fast Chip Select
Dynamic Paging Mode

You say DPM is enabled by default but I find it disabled, at least after "Load Optimized Defaults". All are set to Auto.

There is another one I don't know about called PSB Parking.

Anyway, the partial PAT is keen. I confirmed same with FCS and CTIAW. I noticed something else about SPD -my memory is ratec at CAS2.5 and it seemed to default to 2.5-4-4-8 with F6 but after F7 the greyed out default settings are 2-3-3-7. It runs fine at 2.5-3-3-6 and I am wondering whether to try CAS2 especially as the memory is running a bit under spec frequency.

tallman45, this Gigabyte board has ICH5 not ICH5R.

 

[brainwave]

OK, finally done with my neighbor's Gateway, which btw was infested with the "loveletter" virus and required complete reformat and reinstall of everything....took hours, but he's a good neighbor.

Anyway, thanks for the info Auric. I'm glad you were able to install F7d. Which memory do you have? I would go ahead and try for CAS 2.0 since it will enhance performance on this mobo. But if you oc, you may need to ease up all mem timings. Remember with DDR, FSB frequency is much more important to performance than memory timings.
You also may want to try increasing tRAS to 7 or even 8. Here's a link to a good explanation for this:
http://www.mushkin.com/mushkin/pop-up/latencies.htm
tRAS

About the Advanced chipset settings: On my board when DPM is set to auto, it shows up as enabled. And FCS shows up as disabled. I did not select "Load Optimized defaults." anyway, I just set all three to enabled and everything is ok, or more than ok since I get partial PAT enabled.

I am not familiar with FCS. What is that?

 

[brainwave]

OK, ignore my last question. FCS is fast chip select. I knew that.

I just listened to my own advice and changed my mem timings from 2-3-3-6 to 2-3-3-7. My memory bw increased by about 70 points in SANDRA.

 

[Auric]

Hmm, intelesting. Maybe the DPM is dependent on the RAM used. Mine is Kingston ValueRAM 2.5 (Winbond CH-6).
It cannot be forcibly enabled like the other two settings. Auto results in Disabled. Unless loading the optimized preset has something to do with it but I doubt it.

Okay I did some testing betwixt:
2.0-3-3-6
2.5-3-3-6
2.0-3-3-8
2.5-3-3-8

using the average of a few benchmarks per settings and found 2.0-3-3-6 was the fastest. We aren't talking huge differences here, the largest between samples being 3%. However, I suppose the difference between default 2.5-4-4-8 sans partial PAT must be significant. I will have to try that out. If you did not take the average of several then you might want to recheck the effect of tRAS.

 

[Auric]

Zoinks!

Nevermind what I said before about the greyed out SPD numbers -they do change oddly when enabling/disabling manual settings but after reboot and back into CMOS the greyed numbers display correctly -not that it really matters as they have always been confirmed set correct in Windows.

But back to the zoinks... I did some more testing with default/SPD timings of 2.5-4-4-8. DPM remains Disabled so I cannot check that but the CPC results in a performance increase of 1% on its own, FCS (Partial PAT) 2% on its own and PSB (ready for the zoinks?) 8% on its own!

Now for the difference betwixt 2.5-4-4-8 with all advanced settings disabled and 2.0-3-3-6 with PSB, CPC and FCS enabled... 15%! So I think it really has been worth the messin' and am glad we hashed this out. Hopefully it will help some others too. Keep in mind these settings have not yet been thorougly tested for stability but so far so good

 

[brainwave]

OK, thanks, I'll give that a try and do some more benchmarks.
(Just tell me again what PSB is? I'm sorry and I probably should know but it's escaping me at the moment.....)