2GB Mushkin XP4000 Boot Problem

[Capt Proton]

Anyone using a Newark and Mushkin Redline XP4000? Actually, any socket 754 with 2 Gbs of Infineon Ce-6 will do (this ram, OCZ 4000 Plat, Gigaram 2GB Dual Channel PC-4200, Crucial Ballistix CLIII5N.32 PN56278) Looking for help to get both sticks to boot. Have tried loose timings, different slots, and up to 2.8 VDimm. If anyone is running 2Gbs of Infineon Ce-6, I'd appreciate seeing your timings. Either reply here or PM me.

[Maarek_Stele]

I don't have that memory, but If you look around you'll see that some newark users have problems with ce-6 (clawhammers don't have them to such extent, but still they can't overclock a lot over the 220-230 range with bad timmings). I recall some users having problems to run them over 180mhz on newarks.

[New-B]

Exactly Maarek_Stele.... alot of us Newark users can't boot above 180-185mhz using 2gig of Infineon mem. But there is a workaround using A64tweaker to get it up to speed after getting into windows.

http://www.dfi-street.com/forum/showpost.p...74&postcount=13

[learners permit]

I got 2 gigs to run 234 on this board with the 4000 in and the timings were decent 2.5,3,2,8.

[Capt Proton]

I got 2 gigs to run 234 on this board with the 4000 in and the timings were decent 2.5,3,2,8.

So, you're saying you had this memory working, the new Redline with Infineon chips? If so, do you remember anything about the timings and set up that might be a bit different from the ordinary that may help?

[sailindawg]

Proton, I'm running Infineon CE-6 chips with my A64 3700 Claw. See my sig for 24/7 settings. I'm running the CE-6 on a 5/6 divider @ 220 FSB @ 2,8V, with 2.0, 2,3,5 timings. I can max the RAM around 227 FSB on the RAM, but I'm also maxing out my cpu on a 10x multi at that point and I have to use higher Vcore to maintain stability. I'd rather run the cpu on slightly lower clocks with slightly less voltage to keep things cooler.

 

I can't get my board to run the 2x1G sticks at their stock 250 RAM FSB. I don't think that the NF3 chipset and S754 cpu mem controller combo can handle the CE-6 RAM at it's stock FSB, even at 2T. This board was basically designed around 2x512 TCCD RAM. Generally, 2x1024 really seems to tax this board's chipset and the

S754's mem controller. Some have posted that with 2x1024 UCCC RAM, looser timings, that RAM FSB 250 was achieved. However, with the CE-6, I'd be surprised if your Newark will get much past 220'ish - 230'ish RAM FSB. From what I have read, Newark cores are pretty particular with 2x1024 sticks RAM. LP seems to be doing much better with his Newark on the NF4x with 2x1024. However, it's not all bad news with the CE-6 ! You can pull some really tight timings running a divider. With the high cpu cycles of the Newark and the tight timings of the CE-6, you ought to have pretty darn good gaming machine going !

 

I hope this helps.

[Capt Proton]

I don't think that the NF3 chipset and S754 cpu mem controller combo can handle the CE-6 RAM at it's stock FSB, even at 2T. This board was basically designed around 2x512 TCCD RAM. Generally, 2x1024 really seems to tax this board's chipset and the

S754's mem controller.

Not sure that the Chipset is relevant. From what I understand, and from what Angry_Games has stated, it's all in the CPU. As you are aware, communication between CPU and Memory is direct, it doesn't go through the chipset, so I don't see how the chipset matters that much. But, that's just my understanding.

However, it's not all bad news with the CE-6 ! You can pull some really tight timings running a divider. With the high cpu cycles of the Newark and the tight timings of the CE-6, you ought to have pretty darn good gaming machine going !

I'm afraid that using dividers won't give me the speed I want, as I only overclock 13X230 to 13X235 or so. I have a hard drive in SAA 1, so I have to either stay below 240 or diconnect my hard drive.

Thanks for taking the time to respond to my plea.

[sailindawg]

I have a hard drive in SAA 1

 

That's too bad because that's limiting you right there. If you can stay off the SATA headers 1&2 and turn them off, you will find you will overclock much better.

[Capt Proton]

Roger that, but i want a drive to back up my data to, I want to use RADI 0, and I like having one less bulky cable in my rig, so, there you go. I want my cake and eat it to. I had a 3700 Claw before I got hte Newark. What a difference. Compared to the Newark, the Claw is such a pig. However, it would probably run this RAM. Beauty, as they say, is in the eye of the beholder!

[sailindawg]

i want a drive to back up my data to

 

Gotcha, I use a USB enclosure for a ATA 100 drive for ghosting to. Not the fastest, but I can back up an image without the extra cables inside the case.

 

Compared to the Newark, the Claw is such a pig. However, it would probably run this RAM. Beauty, as they say, is in the eye of the beholder!

 

LOL.....that's one of the reasons that I have stayed away from the Newark 4000. It's pickiness with 2x1024 RAM.

 

Not sure that the Chipset is relevant.

 

My understanding of the chipsets role is that it controls/limits bandwidth. The NF3 chipset is only single channel and my CE-6 chips have a much higher rated bandwidth. At 220, I have maxed out the mem bandwidth on the NF3 chipset. On the NF4 chipset, which allows for dual channel RAM, one can run my CE-6 chips at stock FSB (250) to ~260 stably. At this point, I would assume, that the RAM is now at full bandwidth and the NF4 chipset can handle that bandwidth.

 

So, that's why I said the combo S754 mem controller and the NF3 chipset hold me back on my RAM speed and I have to use a divider. Now, the Newark is a Rev. E mem controller, however, it's on a NF3 chipset. What Learner's Permit has found on his NF4x board, is that his Newark does ~20-30 MHz better with his 2x1024 CE-5 chips (?) with the NF4x chipset than on the NF3 chipset.

 

That's how I understand it. If I'm wrong or not quite understanding it completely, somebody please explain it more fully.

[learners permit]

It seems to me that some changes regarding the circuit board traces to the memory modules was done on this board thus eliminating crosstalk between the modules and their relative data paths. I think most people are aware of the fact that any conductor path creates a magnetic field when current flows through it and the engineers dealt with this issue better on this chipset revision.