Your PC5000=1G?
Printable View
Your PC5000=1G?
YesQuote:
Originally Posted by MikalCarbine
Quote:
Originally Posted by Haltech
My same CPUs do 315+ Mhz 1:1 with TCC5 OCZ PC4800...
Yet with 2GB of PC4000 EB Im cannot break 260 on the memory and with 2GB of PC4000 EL I cant break 274 Mhz. It's the memory controller / 1GB sticks / maybe even a little of the ICs - in any case we are seeing issues with 1GB sticks that even great overclocking CPUs are having issues with.
Radelon - what did you do to get to 282 3d / gaming stable? I can get these EB sticks to 282 Mhz and memtest but I cannot game or 3d bench over 260.
The other item we all seem to be forgetting is that not all sticks are going to do the same speeds / timings....nothing in overclocking is guaranteed...not your CPU, not your motherboard...not your ram.
Haltech, if your unhapy...email me....we will get you sorted out.
Thanks :)
Is yours an E6 atepping? OCZ isnt using the same IC's as the Ballistix, correct?
No they are not the same as ballistix....Micron is only letting those out to Crucial that I know of.
I don't even think I can run 3D at 260 :confused:
Try 259 mhz, a few of us hit that cap with sandy cpus. Max 3d bench was 259, 260 - but I can memtest stable at 282 Mhz.Quote:
Originally Posted by MikalCarbine
I should've mentioned that earlier. Memtest / Prime 32M was fine at 282 2.6v, but 3d bench / game stable was a WHOLE 'nother story - 260 max for me in 3d.
Was it Windows memtest stable above 260? If so, that's REALLY odd...
yea, memtest was fine to 282 @ 2.6v, I was loving it until I started to try and game with it, then I realized 282 wasn't stable in 3d apps and started working back till I found 3d stablity. Was strange but the sticks still rock.
My CPU wont OC as high with 2 x 1GB sticks. The memory controllers are having a hard time with 1GB sticks of memory when overclocking.
Ya it wasQuote:
Originally Posted by Vapor
My Venice won't do 2720 reliably so here is 272 with 9 multi....but I run 265 x 10 all day long for BF2.
http://img455.imageshack.us/img455/7...d2725ec.th.png
i have a set of these on order to replace my Crucial Ballistix 2X1GB. Ive been struggling to get stability with the Ballistix with my X2 @ 260X10 3-4-4-10. I have to put 2.9v in them to stay stable, so if im getting more stability with vdimm, then it must be my ram and not the memory controller. All i hope for with these is stock timings at 260X10.
Anyone have any experience with this ram and an opteron?
I was just getting ready to hit that huge opteron thread. If i see any relevance, ill post a link in reference.Quote:
Originally Posted by MikalCarbine
Actually, 3d benching I can do all the way up to 280.....albeit anything over 274 actually returns lesser scores. The main thing you need to do is to try different bios', I actually checked mine and mines dated 7/06. I played with these sticks upwards, downwards and sideways. They are some of the easiest sticks to clock and to get running but need a bit of tweaking to do their best. The funniest thing about them is, I don't think my Venice tops them out.... for example 2.5-3-2-5, 256 is stable as can be in everything you throw at it, but 257 won't even boot. Same with 282, 3-3-2-7 will work everything but 3d, but 283 won't even boot. You just need to find the best settings for your setup. Mine actually hate looser timings and won't even run well with them. It's all about tweaking, and tweaking starts at what you believe to be your highest clocks. Try 1 setting at a time, then check memtest free of errors. You want the tightest timings you can without loosing stability, most of these timings are the supplemental timings, not the main 4. As you find the absolute best combination at a speed, try to up the clocks, this will give you a "sweet spot". SPD will not find this spot for you because it'll default to looser timings. Check out my thread which has bundles of timings in there, try those for starters and tweak from there. Usually to find best timings per clock takes me about 45 minutes the first time and next to no time tweaking from there. It seems so many are worried about FSB as apposed to best settings. A well tweaked set at 255 will beat a non tweaked set at 280 all day. And at 255 they are guaranteed stable.
What we all need to understand is we are overclocking 2gb worth of ram. Take GSkill for example....arguably the best tccd around....but in the 2x256 fashion. That's only 1/4 of the ram we are working with yet they had so much success because rarely did a memory controller limit the overclock. Now go to the 2x512 market which almost every other company selling TCCD was doing. Higher overclocks are less common and more dependant on the memory controller. If you only got 295 on 2x512 as compared to 315 on 2x256 everyone was complaining. Nobody really ever realised that if you took your 2x512 and used the chips in a 2x256 kit they would probably do 315, and you have to realise this is only 1/2 of what we are trying to overclock. 20 mhz difference in 1/4 the ram in comparison to 1/2 the ram. Now we are talking pushing over 270 in 2x1gb sticks and 4 times the strain on the cpu. By my calculations anything over 255 is icing on the cake for 2x1gb, but yet people aren't happy. And these limitations are all a result of the memory controller. The people with the highest TCCD clocks are the people that usually have the best cpu's (FX series) and we all think our little half wit cpus should be able to do the same. There's a reason why the FX series is what it is, the cream of the crop. Venice is usually better than a San Diego (non fx) because the venice is made to be at 512k on cache and there's no FX with that spec, so some venices have great memory controllers. Whereas San Diego's are basically the FX's that aren't the cream of the crop, therefore have weaker memory controllers. Not to mention the fact that some motherboard chipsets are better than others. So unless you have umteen boards and umteen cpus that can prove that the ram isn't performing the way it should then you need to take into consideration those as the weaker link before pointing the finger at the ram. Jumping to conclusions will only drive the prices up.
And as for dual core cpu's.....inherently I would say you are only as good as the weaker of the 2 cores. Therefore, more problems with memory controllers will be shown and in most cases will not be desirable.
Things are the way they are and we can only find ways to work around them. If you don't have the money to hang with the big guns around here, don't expect the same numbers from a cpu that costs 1/4 of the price. If it ain't the real thing then it's probably not the best. We can all argue that I'm just speculating, but from my experience, those having problems with 2x512 or 2x1gb ram can send them to someone not having problems and voila problem solved....but wait....is it? Not really, because all we found is that the ram wasn't the problem and where do we go next.... a different brand? .... a downgrade to 2x256 because your memory controller can handle it? Fact is you still didn't fix the problem all you are doing in these cases is avoiding the problem. And avoiding the problem only covers it up until it resurfaces again.
Sorry for the long post but I guess sometimes the table needs to be set before people actually think that dinner is ready.
Why am I happy to move from Mushkin back to OCZ?
Answer is bellow.
Board: Ultra-D
Bios: 623-3
Slots: Yellow
In DRAM Configuration Section of Genie BIOS:
200
enable
3.0
03
08
02
8
14
02
02
02
03
4708
auto
enabled
auto
0
level 6
auto
auto
DRAM Response Time: Fast
auto
256
disable
16
07
disable
In Genie Bios:
285
auto
16 16
7
100
disable
1.500
1.30v
above VID * 123
1.30v
1.70v
2.8v
Disable
:yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes:
http://img473.imageshack.us/img473/7...superpi9fy.jpg
:yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes: :yikes:
I had to lower the multi to 7 cause my memory controler could not take 2Gb @ 285x9.
One of the sticks does 290 htt prime stable but the "bad" one is stuck @ 285 so this will be the most I can get from them.
If I could loose some of the timmings I might reach 300 htt. Anyone has any ideas?
Can u 3d bench them at that speed though?
Try 3dmark 2001 se or see if you can game with them that high?
Those are the EB not the ELs?
This are
OCZ EB DDR PC-4000 2x1024MB Platinum Edition Dual Channel Kit
I will test 3d mark 2005 as I get home tonight.
Quote:
Originally Posted by Radelon
Well put and I whole heartedly agree Radelon. :toast:
sure will soon... playing with 2x Opteron 144 CABYE 0536 GPMW right now figuring them out, then have 2x Opteron 146 CABYE 0540 FPBW coming todayQuote:
Originally Posted by MikalCarbine
will whip in my 2x1GB OCZ PC4000EB sticks once i figure the cpus out :)
so far opteron 144 mem controllers are pretty similar to my FX-57 0516 WPMW - but still not as good as FX-57 mem controller clocks about same but ram such as TCCD and BH-5 need slightly loose advance timings on DFI NF4 UT SLI-DR compared to FX-57
but still opteron 144 mem controller is better than 3700/4000+ sandies i tested and way better than mem controller on venice/winchester cpus for winbond BH-5 and TCCD ram at least
with opteron 144's might have to see how 2GB OCZ PC4000EB handles dividers due to low 9x cpu multiplier :)
Guys I just need to make a point about our kits and the guys who i have posting info for you.
I instructed all our guys to prove the modules are 3d stable, this is VERY IMPORTANT as the modules will run super pi a lot higher than they will 3d benches.
So if you are looking to compare products please make sure you see 3d stable and not just a 32m pi screenie...that is not a sure way to sshow they are stable.
As mentioned in this thread a 2gig kit is quite a load on the memory controller and 3d does seem a lot more stressfull than Pi.
I wub you :DQuote:
Originally Posted by eva2000
Okay, here is my 3D Mark 05
http://img344.imageshack.us/img344/4...mark2759ag.jpg I had to lower from 285 to 275 so my CPU can handle.
Still a DAMN good stable 3D clock