Corsair 800ul vs 8500

[PFOUR]

Try more MCH voltage, could be the board not the ram thats failing.

[Steensen]

Try more MCH voltage, could be the board not the ram thats failing.

Thx for the advice, but they still wan't do 1066 when I set it for DDR800 devider.
MCH is at 1.55 currently, and also when I tried out at DDR800 devider (1:2), at that speed, the MoBo can't be the problem.

[Redbeard]

Thx for the advice, but they still wan't do 1066 when I set it for DDR800 devider.
MCH is at 1.55 currently, and also when I tried out at DDR800 devider (1:2), at that speed, the MoBo can't be the problem.

Wow, you guys are on top of things in here. If you're having problems running 2 modules at 1066 at 5-5-5-15 at 2.2V, we'll RMA them no problem.

1066+ MHz frequency is pretty high and a lot of the boards lately that we've been testing in the lab have some issues with 4x512 but work just fine with 2x512. Also, we've noticed that for whatever reason the black slots work a little better for our internal testing than the yellow slots, which we're working on with Asus.

As for top end, we see a lot more headroom in the new Microns than the old fatbodies, and the old fatbodies were awesome. I'm sure you've got active cooling on them and on the northbridge, so I won't even point that out.

Seriously though if two of the modules fail at 1066 @ 5-5-5-15 @ 2.2VDIMM, send them back to us and we'll replace them with something faster.

[boostedevo]

boostedevo That's what I thougt too, but yet again, the Dimm mosfet can be a bottleneck, and not capable of supplying enough juice to the dimm's, and therefore they cannot run the specs.
Try out two dimms this weekend I think, otherwise, I'm going to RMA the dimms (all four of them), to see if that helps.
Just weird that the dims runs super stable at 1000, but 1010, then they want complete a memtest86.

Redbeard posted a possible issue on slots. If you have a multimeter and think the board might be the issue, I'd measure the DIMM fet or better yet, the memory side of the coil, and see how that compares to your BIOS setting for Vddr. If it's .1-.2V lower, that could be a problem and it would be the first I've heard of anything like that. My unmodded board was obviously a little lower than set, but just a couple hundreths of a volt.

As for top end, we see a lot more headroom in the new Microns than the old fatbodies

Really?? As in the current ICs in new 8500s or a new batch coming in? Is the CAS4 issue (ie. none) just from programming?

and the old fatbodies were awesome

Thats the fact Jack! I was REALLY hoping the new ICs would cough up a solid DDR2-1100+ @ CAS4, even if it took a little more voltage. See anything like that in the near future?

Corsair is always on the ball with service...to see you pop up in the forums on you own accord and offer to handle Steenson's issue is just awesome!

[AbsoluteZero]

These chips are owning. Mobo is P5WD2-E.. Mobo has vdd2 mod but rams are only at 2.3v max.. I'll push further later this week..

[boostedevo]

They're not owning yet, but damn close Congrats! Impressive for 2.30V...did you measure it on the board to verify you didn't accidentally leave the pot turned up because that's a complete 180 from the 8500s I've seen so far? Or are those 8000ULs...you didn't specify That's like D9 type speeds/timings with hardly any voltage. I'll be curious how high you can carry the CAS4 with voltage. If it's the 8500s and you can get close to DDR1200 CAS4, I'll still have hope for them Good luck and keep up posted!

[Steensen]

Wow, you guys are on top of things in here. If you're having problems running 2 modules at 1066 at 5-5-5-15 at 2.2V, we'll RMA them no problem.

1066+ MHz frequency is pretty high and a lot of the boards lately that we've been testing in the lab have some issues with 4x512 but work just fine with 2x512. Also, we've noticed that for whatever reason the black slots work a little better for our internal testing than the yellow slots, which we're working on with Asus.

As for top end, we see a lot more headroom in the new Microns than the old fatbodies, and the old fatbodies were awesome. I'm sure you've got active cooling on them and on the northbridge, so I won't even point that out.

Seriously though if two of the modules fail at 1066 @ 5-5-5-15 @ 2.2VDIMM, send them back to us and we'll replace them with something faster.

See my sig. Then you see I've got water on most of my setup, including NB (That sits first in the NB/GFX/HDD/Mosfet loop), so temp on NB is no prob.
The RAM are cooled by a 12db Papst 80mm, so they are 1-3°C above room temp that's laying around ~23°C.
Mounted G-RAM cooling profiles on the mosfets not cooled by water, and the Dimm mosfet doesn't seem to become warm at all, so it should be wierd if the dimm mosfet is the problem.

Wich Mobo did you have problems with 4*512MB in?

Nice to know about the black vs. yellow slots. Thx m8.

If I get anny problems with 2*512MB, then I'm gonna quote you to my dealer, so that i can/will send the dimm'sfor RMA.

Redbeard posted a possible issue on slots. If you have a multimeter and think the board might be the issue, I'd measure the DIMM fet or better yet, the memory side of the coil, and see how that compares to your BIOS setting for Vddr. If it's .1-.2V lower, that could be a problem and it would be the first I've heard of anything like that. My unmodded board was obviously a little lower than set, but just a couple hundreths of a volt.

Where is that messure point? (please show on a pic.)

[boostedevo]

This is on my P5WD2 *Premium*...not sure if the -E is the same. Ignore the stuff in red, it's old pic from when I was modding it. Measure either the center leg on that fet or that solder point on the coil. The coil is better (would show least voltage) and safer...plenty of people have slipped and shorted things trying to measure those short legs on fets

[Steensen]

THX. Looking into it later.Work calls.

[proth]

Lets see the Benchmarks for both these chips so we can actually compare.

I'm really not interested in benching anything now that have to run under 5GHz with the warm weather and I don't usually don't post any lower scores that posted before. Someday my Chilly1 SS will be here, hopefully before Conroe hits the general public.

If you want me to bench anything in particular for you to compare, let me know what it is and I'll gladly do it for you.

Ran memtest on the 8500 and didn't get them stable to around 1050 @ 5-5-5. Tried each stick seperately and they are about the same.
Weird thing is they only run 5-5-5. Can't even run 4-4-4 @ 1000
Also did a Vdimm mod and running them up to 2.7 doesn't help. Looks like Absolute is getting there...

[Steensen]

Proth Sad story. Gonna test mine later today.

[AbsoluteZero]

They're not owning yet, but damn close Congrats! Impressive for 2.30V...did you measure it on the board to verify you didn't accidentally leave the pot turned up because that's a complete 180 from the 8500s I've seen so far? Or are those 8000ULs...you didn't specify That's like D9 type speeds/timings with hardly any voltage. I'll be curious how high you can carry the CAS4 with voltage. If it's the 8500s and you can get close to DDR1200 CAS4, I'll still have hope for them Good luck and keep up posted!

When I mod the board I only solder a wire, I put the pot on a breadboard and complete the circuit from it with wire. The pot was not connected. I've 2 boards. P5WD2-E and P5WD2 ..-E can give max 2.4v from bios but other one gives only 2.3 .. I always used my 5400UL with P5WD2, to compare both of them, I used only 2.3v from bios. Will put some more tests later..

[Redbeard]

Redbeard posted a possible issue on slots. If you have a multimeter and think the board might be the issue, I'd measure the DIMM fet or better yet, the memory side of the coil, and see how that compares to your BIOS setting for Vddr. If it's .1-.2V lower, that could be a problem and it would be the first I've heard of anything like that. My unmodded board was obviously a little lower than set, but just a couple hundreths of a volt.

Really?? As in the current ICs in new 8500s or a new batch coming in? Is the CAS4 issue (ie. none) just from programming?

I'm not familiar with the CAS4 issue you're mentioning. We've got them running in the labs at CAS4 at 1000+, sometimes higher, though I can't really give out too many details on that. My suggestions with these ICs for best performance so far:
-Active cooling, these things get hot, especially with 2.4V (or more, if you're a crazy voltmodding lunatic like some of you guys, I hope you're wearing goggles for when these things burst into a billion pieces from heat)
-High voltage
-Keep an eye on tRCD, unlike the older fatbodies, where you could run 3-2-2 almost as fast as 3-3-2, these ones are much more susceptible to the tRCD being the limit. We see 4-5-4-12 hit almost as high as 5-5-5-12 with the same voltage. If I was trying to hit 1100, I'd be at 4-5-4 and 2.4 (or higher) volts.

We use the P5WD2-E board with a 955 processor and air cooling on the northbridge, 0401 BIOS, and we test everything internally at 2.2V. We also use RST Pro2 cards for stability testing, which correlate pretty well to Windows stability.

If you want shots of our BIOS settings for our internal testing I can take those later, but I've got lunch to eat right now.

[Steensen]

I've just been testing My dimm's, and 2*512 ran 1066 5-5-5-15 2.2Vdimm through memtest (diden't wanna go through heavy stability testing), without a hitch.
CAS 4 was possible with 2.4Vdimm 1066, but at 1080, they maide errors in memtest.

As soon as I put in the second pair of dimms, no matter witch pair sits in witch slots, My dimms will not run 1066 5-5-5-15 at 2.4V, so neather at 2.2V.

My MoBo is not moddet, except for watercooling on NB and mosfets, and cooling profiles on most other mosfets (those in between the capacitors, between edge of MoBo and socket, these get quite hot at full CPU load).

All these test has been done with the 0101 BIOS, so I guess, that it's time to flash to 0401.

Has anyone tested the 0422?


Redbeard Have you been able to get 4*512MB 8500's to run 1066 in your lab P5WD2-E MoBo, with the 955EE?

[Redbeard]

Redbeard Have you been able to get 4*512MB 8500's to run 1066 in your lab P5WD2-E MoBo, with the 955EE?

We haven't tested 4x512 successfully yet as far as I know, we qual things in pairs right now, but I'll throw some on a board and respond with results tomorrow.

[mike]

Wow, you guys are on top of things in here. If you're having problems running 2 modules at 1066 at 5-5-5-15 at 2.2V, we'll RMA them no problem.

1066+ MHz frequency is pretty high and a lot of the boards lately that we've been testing in the lab have some issues with 4x512 but work just fine with 2x512. Also, we've noticed that for whatever reason the black slots work a little better for our internal testing than the yellow slots, which we're working on with Asus.

As for top end, we see a lot more headroom in the new Microns than the old fatbodies, and the old fatbodies were awesome. I'm sure you've got active cooling on them and on the northbridge, so I won't even point that out.

Seriously though if two of the modules fail at 1066 @ 5-5-5-15 @ 2.2VDIMM, send them back to us and we'll replace them with something faster.

[proth]

Thanks Redbeard. I'm waiting for an RMA number as we speak. It's great customer support like this as well as a great products that keep me dedicated to Corsair.

[AbsoluteZero]

I'm not familiar with the CAS4 issue you're mentioning. We've got them running in the labs at CAS4 at 1000+, sometimes higher, though I can't really give out too many details on that. My suggestions with these ICs for best performance so far:
-Active cooling, these things get hot, especially with 2.4V (or more, if you're a crazy voltmodding lunatic like some of you guys, I hope you're wearing goggles for when these things burst into a billion pieces from heat)
-High voltage
-Keep an eye on tRCD, unlike the older fatbodies, where you could run 3-2-2 almost as fast as 3-3-2, these ones are much more susceptible to the tRCD being the limit. We see 4-5-4-12 hit almost as high as 5-5-5-12 with the same voltage. If I was trying to hit 1100, I'd be at 4-5-4 and 2.4 (or higher) volts.

We use the P5WD2-E board with a 955 processor and air cooling on the northbridge, 0401 BIOS, and we test everything internally at 2.2V. We also use RST Pro2 cards for stability testing, which correlate pretty well to Windows stability.

If you want shots of our BIOS settings for our internal testing I can take those later, but I've got lunch to eat right now.

Redhead, my 8500 is just come from corsair. They are very sensitive to 2 timing .. in 3-2-2-x they cant hit as 5400ul or 8000ul but at 3-3-3-3 they made 840 ( 8000ul hit 790 with the same conditions ). 2.3v 4-4-3-4 @ 1100MHZ no active cooling on memory or northbridge. (8000ul hit 1040 at 4-4-4-4 2.3v) .. also i gave up to 2.8v to 8000ul .. I was planning to give up to 2.6v.. But now I'll be more careful about the voltage because of this heat problem.

[boostedevo]

I'm not familiar with the CAS4 issue you're mentioning. We've got them running in the labs at CAS4 at 1000+, sometimes higher, though I can't really give out too many details on that. My suggestions with these ICs for best performance so far:
-Active cooling, these things get hot, especially with 2.4V (or more, if you're a crazy voltmodding lunatic like some of you guys, I hope you're wearing goggles for when these things burst into a billion pieces from heat)
-High voltage
-Keep an eye on tRCD, unlike the older fatbodies, where you could run 3-2-2 almost as fast as 3-3-2, these ones are much more susceptible to the tRCD being the limit. We see 4-5-4-12 hit almost as high as 5-5-5-12 with the same voltage. If I was trying to hit 1100, I'd be at 4-5-4 and 2.4 (or higher) volts.

We use the P5WD2-E board with a 955 processor and air cooling on the northbridge, 0401 BIOS, and we test everything internally at 2.2V. We also use RST Pro2 cards for stability testing, which correlate pretty well to Windows stability.

If you want shots of our BIOS settings for our internal testing I can take those later, but I've got lunch to eat right now.

Just the fact that proth wasn't getting CAS4 at lesser than rated speed and with increased Vddr and that even Steensen was having trouble running at rated specs. Looking at Absolute's results in comparison, I thought for a second he was talking about 8000ULs and not 8500s His results and your info gives me new hope. It would take a lot to convince me to give up the D9s altogether, but things seem to be looking up now.

2.3v 4-4-3-4 @ 1100MHZ no active cooling on memory or northbridge. (8000ul hit 1040 at 4-4-4-4 2.3v) .. also i gave up to 2.8v to 8000ul .. I was planning to give up to 2.6v.. But now I'll be more careful about the voltage because of this heat problem.

Excellent results Absolute! Is that 4-4-3-4 @ 1100 stable/error-free on tests? That would be impressive on 2.3V. If you can pretty much keep those timings @ 1200 with upped Vddr, I may have to give them a shot. Nice idea with the breadboard for the mod BTW

[AbsoluteZero]

Excellent results Absolute! Is that 4-4-3-4 @ 1100 stable/error-free on tests? That would be impressive on 2.3V. If you can pretty much keep those timings @ 1200 with upped Vddr, I may have to give them a shot. Nice idea with the breadboard for the mod BTW

It finished pcmark02 .. You can see screenshots a few posts above in this topic.

[boostedevo]

Yes, I am a dolt, LOL. I was thinking along the lines of memtest, but any benches a little newer or tougher coming up for the next round of testing? PCMark04 is much tougher to pass with a score than 02 and SPi 32M would test them pretty well. Close to the limit, I can pass SPi 1M at speeds/timings that will error after 1 iteration of 32M. If it's really on the hairy edge, it will pass 1M and not even 4M

Thanks again...looking forward to your results...

[nFo]

I want some new ram, sould I go for: 8000UL or 8500 ?
When I look at the results so far there are close, but 8000UL has better timings.
And what's the max or min voltage of Vddr ?

[Harshal]

^^^ I know...very confusing man. But I really hope to get what Abolute 0's performance is, if I get those PC-8500 I mean 540+@4-4-3-4 @ 2.3V is really super clocks.

[proth]

8000ul is End of Life and replaced by 8500, Doubt if you can find them anywhere.

[Harshal]

^^^ Too bad no 2GB modules with any