Socket 939 Discussion Thread

[Guest LithoTech]

I found mine didn't like special vid, but at this point it's worth a shot. Puts it a squeak over 1.55v. I noticed your LTD and Chipset are at default as well, I'll give that a try too.

 

I had to give up on it. I ran into the same instability I usually get when using the special VID control. Once I had enough voltage to stop P95 from locking up, she'd simply error within 4-6 tests at about 15-17 minutes.

 

I pushed it all the way to 1.6vcore with no improvement on P95 errors. The lockups stopped way before 1.6volts, so I felt going beyond 1.6v would not help since it wasn't already helping.

 

So the X2-3800 is in, and the Opty resting on a shelf while I decide what to do with it: HTPC for livingroom, using a DFI RS482, or slap it into some mobo, a spare case and other parts, and sell it as a complete system.

 

I'll be working on overclocks for the 3800 over the next couple of weeks. First I'll get it to 2.7ghz to match my other entries for comparrisons. Then I'll let the reins out and see what she'll do.

 

Note: I noticed that this X2-3800 gets worse memory scores than the Opty did using the same ram and timings, and benching with Everest. Particularly, a much higher latency score. I am assuming this is due to the memory controller having to deal with two cores -- would this be correct? Or is the Opty really that much better at handling the memory?

 

.

[Anakin_hollan]

Looks good now!

 

2) Just notice you have another 7800GT that is underperforming, the 3DM01 score is below my 6800GS just like Anakin's. You underclocking it as well?

 

Well, the setup says:

 

eVGA 7800GT, 490/1200 81.98 drivers

 

But the 3dmark03-score is quite low as well (10609) while 05 is higher than me? Can't be the CPU, he's runinng faster than me. Maybe driver-issue here? 81.98 is quite old? On the other hand, it's the same version as you're running... :confused:

 

About the latency, I saw the same difference. At 258Mhz 1:1 I had 43ns on the Opty, and 51 on the X2.

[sailindawg]

It's interesting to see that the last few posts in the OCDB, besides mine, are using Mushkin Redline EP PC4000, Infineon CE-6 chips. I got mine in January when I was running my NF3 board. That S754 cpu just couldn't do those RAM justice. However, what I really like about the Redlines, is that I was able to run very tight timings @ 225FSB on the NF3 board. Even with my San Diego, my RAM timings are pretty tight at 265 FSB ! Good to see more people getting very good results with these RAMs.

[madchild]

Looks good now!

 

Couple of things...

 

1) One minor typo:

 

 

 

Multi should be 11x

 

2) Just notice you have another 7800GT that is underperforming, the 3DM01 score is below my 6800GS just like Anakin's. You underclocking it as well?

 

3) CPU stepping: I thought you might have been the user that needs to pull his XP-120 to get the stepping, but that was Viking. Do you have your stepping recorded anywhere? If so, just edit it in.

 

Good job! Looking forward to seeing what your stepping is on that 4200. The system I just built for my brother has an LDBFE 0622 FPMW. We haven't overclocked her yet, still teaching him the basics and working out some minor bugs he still has. But I'm always curious as to what his stepping may be capable of, including what other steppings are doing.

 

1) Typo fixed.

 

2) Just realized that I was still running really old drivers... just updated to 91.31 running new benchies now..

 

3) LDBFE 0612 FPMW

 

Im still pushing it to go further. I can boot up at 2.8 but it fails prime on core1 after about 4 hours... wonder if I should push a bit more voltage.. Im new to these X2's so Im not sure how far I should go... my temps are good, so maybe 1.55? Im also thinking about playing around with some newer bios, but Im not sure which to go with...

[Roger_D25]

It's interesting to see that the last few posts in the OCDB, besides mine, are using Mushkin Redline EP PC4000, Infineon CE-6 chips. I got mine in January when I was running my NF3 board. That S754 cpu just couldn't do those RAM justice. However, what I really like about the Redlines, is that I was able to run very tight timings @ 225FSB on the NF3 board. Even with my San Diego, my RAM timings are pretty tight at 265 FSB ! Good to see more people getting very good results with these RAMs.

 

Hey Sailindawg,

Yeah this RAM is pretty impressive stuff, I was lucky because they speed-binned my modules before they were shipped due to my luck with defective Redlines (UTT) over the course of the last year. They went through a handful of kits using their RST Pro2 tester as well as an nf4 DFI SLI-DR motherboard and ended up with a kit stable at 285Mhz @ 3-3-2-8 1T on 2.7v. With that said though I was also told that these modules experience a 3D phenomonon which makes the modules have a somewhat lower 3D stability, should be 3D stable upwards of 280ish give or take.

 

Also after getting my sticks I pulled the heatspread off because right off I noticed they had a green PCB, not black (I was worried they might be Mushkins HP line, not the XP)! After pulling the spreaders I also noticed these modules are running CE-5 IC's instead of CE-6 IC's. Since I'm no expert with 2GB sticks I wrote back to Mushkin, they replied,

 

"We have used both black and green PCBs on these. We try to use black whenever possible but sometimes this creates lead-time issues and if we are in a hurry we will just build on green. Black looks better I know and we are trying to get to a point where we are better prepared for demand spikes by carrying more black PCBs in inventory. We are also going to be implementing black PCBs on our DDR-2 redline in the near future.When the 991492/991493 was originally released we used CE-6 chips because CE-5 was not readily available at the time. Then, CE-5s were available in enough quantities to support our demand and we have been using the -5 for at least ½ a year now. In case you don’t know, the -5 and -6 are nanosecond ratings; -5 is 200MHz (PC3200), -6 is 166MHz (2700). For these particular chips the ns rating doesn’t really matter as the top end will usually be about the same. The die rev is important (the CE portion). The full chip number is HYB25D512800CE-(5/6) if you want to find further information. Your modules are the -5 version."

 

After hearing back I felt much better, in hindsight I guess it shouldn't have mattered as both PCB's were Brainpower units. Anyway I have a long ways to go before I get up around 280Mhz but hopefully the'll get somewhat close on my Expert!

 

Hey Lithotech,

I had gone through 3 sets of Redines (UTT) before this last RMA replacment. I'm not sure exactly what the problem was but after time my system would occationally not bootup after shutting down (I got the dreaded 3LED's along with the beeeeeep, beeeeeep....... code! If I waited a bit between shutdown and powerup it would usually fire rigth up. After a few long weeks of this my system just stopped booting with both sticks installed no matter what I did or how long I waited! During this whole time I had been testing voltaged, settings, swithcing RAM slots, different BIOS versions, with no luck at all. Long story short, one stick was fine but the other one was just dead! This was basically the same thing that happened with each of my RMA replacments?

 

One last thing and don't quote me on this! I was asked if I wanted to wait a bit because Mushkin was currenlty binning a small yeild of UTT IC's but they weren't sure how many kits they would come up with? If their anything like the older 1GB Redlines they should be pretty decent!

[likewhoa]

Hey Sailindawg,

Yeah this RAM is pretty impressive stuff, I was lucky because they speed-binned my modules before they were shipped due to my luck with defective Redlines (UTT) over the course of the last year. They went through a handful of kits using their RST Pro2 tester as well as an nf4 DFI SLI-DR motherboard and ended up with a kit stable at 285Mhz @ 3-3-2-8 1T on 2.7v. With that said though I was also told that these modules experience a 3D phenomonon which makes the modules have a somewhat lower 3D stability, should be 3D stable upwards of 280ish give or take.

 

Also after getting my sticks I pulled the heatspread off because right off I noticed they had a green PCB, not black (I was worried they might be Mushkins HP line, not the XP)! After pulling the spreaders I also noticed these modules are running CE-5 IC's instead of CE-6 IC's. Since I'm no expert with 2GB sticks I wrote back to Mushkin, they replied,

 

"We have used both black and green PCBs on these. We try to use black whenever possible but sometimes this creates lead-time issues and if we are in a hurry we will just build on green. Black looks better I know and we are trying to get to a point where we are better prepared for demand spikes by carrying more black PCBs in inventory. We are also going to be implementing black PCBs on our DDR-2 redline in the near future.When the 991492/991493 was originally released we used CE-6 chips because CE-5 was not readily available at the time. Then, CE-5s were available in enough quantities to support our demand and we have been using the -5 for at least ½ a year now. In case you don’t know, the -5 and -6 are nanosecond ratings; -5 is 200MHz (PC3200), -6 is 166MHz (2700). For these particular chips the ns rating doesn’t really matter as the top end will usually be about the same. The die rev is important (the CE portion). The full chip number is HYB25D512800CE-(5/6) if you want to find further information. Your modules are the -5 version."

 

After hearing back I felt much better, in hindsight I guess it shouldn't have mattered as both PCB's were Brainpower units. Anyway I have a long ways to go before I get up around 280Mhz but hopefully the'll get somewhat close on my Expert!

 

Hey Lithotech,

I had gone through 3 sets of Redines (UTT) before this last RMA replacment. I'm not sure exactly what the problem was but after time my system would occationally not bootup after shutting down (I got the dreaded 3LED's along with the beeeeeep, beeeeeep....... code! If I waited a bit between shutdown and powerup it would usually fire rigth up. After a few long weeks of this my system just stopped booting with both sticks installed no matter what I did or how long I waited! During this whole time I had been testing voltaged, settings, swithcing RAM slots, different BIOS versions, with no luck at all. Long story short, one stick was fine but the other one was just dead! This was basically the same thing that happened with each of my RMA replacments?

 

One last thing and don't quote me on this! I was asked if I wanted to wait a bit because Mushkin was currenlty binning a small yeild of UTT IC's but they weren't sure how many kits they would come up with? If their anything like the older 1GB Redlines they should be pretty decent!

 

my Munskins are NOT 3D stable at 346x9 on 166 divider 3-3-2-1 @ 281Mhz. I'm sure I just need to tweak them a bit more or else I will just lower the Divider.

[MacBooom]

lo all,

 

added my first complete benched OC of my sandy 4000

 

this post, can you verify its correctly posted?

[likewhoa]

looks good to me

gj

[Praz]

lo all,

 

added my first complete benched OC of my sandy 4000

 

this post, can you verify its correctly posted?

You might want to go back and edit the bios settings so they look like what everybody else has posted. Everest read should be included in one of the other screenshots. But it's still a nice overclock.

[Roger_D25]

likewhoa,

You should be able to run them atleast at 250Mhz 1:1 as that is what their rated for? Is there any reason your running a divider at only 246Mhz? Last but not least can you post your DRAM BIOS settings so we can help you get this sorted?

[MacBooom]

no i'm not using a devider, running them on 1:1 200

 

have been working on a 12*250 for 2 weeks now, but could not get that one stable in 3dmark03, but it runs 3dmark06 without any problem

 

also tried lowering to fsb 249 and 248, but same results, used in both cases the same memtimings as in the posted in oc DB.

 

but it runs @3ghz fine on games (cod2 dx7 maxed out and bf2) for whole evenings, but it is not a real oc if its not stable @ all tests.

 

i thinking about trying the 11*273 to reach the 3ghz stable, thats my goal

the mem should be stable @ 3448 1T til 275mhz i read in some tests

[likewhoa]

likewhoa,

You should be able to run them atleast at 250Mhz 1:1 as that is what their rated for? Is there any reason your running a divider at only 246Mhz? Last but not least can you post your DRAM BIOS settings so we can help you get this sorted?

 

look at the last OCDB entry I posted,.. 336x9 05/06 divider.

 

I can run it at 1:1 but that is only limited in my quest for higher Cpu Mhz, but now at seems my 05/06 divider has meet it's limit as running the RAM at over 280Mhz is not 3d Stable. Gonna try a few more tweaks before I lower it down to 03/04 divider. but It would have been nice to run at 280Mhz+ on this dual-core opteron 165 I got running prime stable at 346x9