2x1GB/1x1GB OC Database

Printable View

Show 100 post(s) from this thread on one page

07-08-2005, 12:03 PM
AMD-me

Quote:

Originally Posted by gundamit

AMD-Me over at an OCForums thread claims Smasung UCCC in their PC3700 1gig sticks.

All the results for the 2x1024 set-up is 1T. I'm just gaming with those sticks right now. Any time I have for benching is going towards a Venice I just picked up and of course I'll be testing 2x512 to start. It may be a while before I detemine whether the Venice is fast enough to get a spot in my gaming rig. By then I may have elected to go with the Micron since it seems to kick ass. Wish the price would come down a bit ...

Hey I have some more info! Heres a copy of my email

Code:

That model uses either Micron -5BD or Elixir chips. The higher 3700 line uses the Samsung UCCC ---------- Original Message ---------------------------------- From: perek@CLEMSON.EDU Date: Thu, 7 Jul 2005 14:26:32 -0400 (EDT) >> Hi, >> >> I spoke with you on the phone yesterday and promised you some information >> on the module with part number ending in 9053. After speaking to someone >> in our sales department, that model number is now discontinued and is >> actually now known as our GR1DD8T-K2GB/433/2.5 model, which is the Mach >> 1000 PC3500 in 2GB size. The ram timings are 2.5-4-4-7. >> >> Hope this answers your question. >> >> Larry >> >> >> >> >> ________________________________________________________________ >> Sent via the WebMail system at gigaram.com >> >> >> >> >> > >Any idea on what type of chips to expect in it? >Micron -5B D >Samsung UCCC maybe? > >Thanks for the great response, >Pere Kyle
07-08-2005, 02:30 PM
derektm

Whats better.....

Micron -5B D @ high speeds and looser timings or tighter timings and lower speeds??

Thanks
07-08-2005, 02:38 PM
ozzimark

seeing as my sticks top out at:

2.5-2-2 @ 203mhz
2.5-3-2 @ 223mhz (3-3-2, 2.5-3-3 are the same, interestingly enough)
3-3-3 @ 251mhz

i'd say the latter wins :D
07-08-2005, 02:53 PM
Vapor

I'd say high speeds too...
07-08-2005, 03:09 PM
Jericho

Higher with slightly slacker timings is faster overall in SuperPI/3Dmark for me
07-08-2005, 03:31 PM
derektm

Can someone post a Sandra mem bench?
07-08-2005, 04:45 PM
brandinb

guys i have seen many reviews of ocz 2x1gb gold reaching ddr 480-500 with cas 3-3-3 and they are cheap!! 215 at newegg atm. i am suprised nobody has a review of that set as it seems exceptional and will oc much better then the platinum set which only does ddr440 altho at better timings thats still a poor oc. cmon anyone have a try at this set and does anyone know what chips they use in there???
________
LAMBORGHINI V12 HISTORY
07-09-2005, 01:38 AM
Cashman

Here is a Mem Band width shot at 260mhz

http://img250.imageshack.us/img250/5...8mem9to.th.jpg
07-09-2005, 06:19 AM
burningrave101

Quote:

Originally Posted by nugzo

3120 and auto for the drive srength.

Anyone else? I'm using 3120 at the moment as well but a tREF value of 3120 is 200Mhz MEMCLCK 15.6us and 1GB DIMM's aren't going to have a higher refresh rate then 7.8us. I'm surprised their not 3.9us considering how 256MB DIMMs are 15.6us and 512MB are 7.8us.

What about drive strength combinations? Do most of you just leave it to Auto or are you using normal or weak strengths?
07-09-2005, 08:43 AM
nugzo

gonna try 0016 and 3072
07-09-2005, 08:48 AM
nugzo

ok seem to be doin ok @ 3072. what's a good test to tell if i'm getting a performance change?
07-09-2005, 09:03 AM
burningrave101

Quote:

Originally Posted by nugzo

ok seem to be doin ok @ 3072. what's a good test to tell if i'm getting a performance change?

I think the tREF setting is primarily a stability setting more then it is a performance setting. I dont really know how you would test to find out if your using the best tREF setting.
07-09-2005, 09:07 AM
derektm

Quote:

Originally Posted by Cashman

Here is a Mem Band width shot at 260mhz

http://img250.imageshack.us/img250/5...8mem9to.th.jpg

Nice bandwidth...but my question is, at those loose timings doesn't that effect game's and such or does the speed make up for it? Like some 2x1gb RAM is rated 2-3-2-5 @ 200mhz. Do latencies even affect the bandwidth in sandra?

Thanks
07-09-2005, 09:21 AM
nugzo

Quote:

Originally Posted by derektm

Nice bandwidth...but my question is, at those loose timings doesn't that effect game's and such or does the speed make up for it? Like some 2x1gb RAM is rated 2-3-2-5 @ 200mhz. Do latencies even affect the bandwidth in sandra?

Thanks

speed makes up for it
07-09-2005, 09:30 AM
nugzo

0016 tref took a big hit. latency went from 41ns to 55ns on everest. super pi 1m from 29s to 33s.
07-09-2005, 10:05 AM
TransNone13

Cashman, can you take a picture of your BIOS settings under Ram? That would be very helpful :D.
07-09-2005, 11:55 AM
burningrave101

Quote:

Originally Posted by derektm

Nice bandwidth...but my question is, at those loose timings doesn't that effect game's and such or does the speed make up for it? Like some 2x1gb RAM is rated 2-3-2-5 @ 200mhz. Do latencies even affect the bandwidth in sandra?

Thanks

With Athlon 64's the CPU speed has a huge impact on gaming performance while timings and RAM speed only have a small impact. Latency does affect bandwidth, especially the command rate, but theoretical bandwidth doesn't mean much. In real world performance you'll be lucky to ever notice the difference between 200Mhz and 300Mhz RAM and 2-2-2-5 and 3-4-4-8.

Quote:

Originally Posted by nugzo

0016 tref took a big hit. latency went from 41ns to 55ns on everest. super pi 1m from 29s to 33s.

If your running the RAM 1:1 200Mhz divider then try 1560 and 0780.
07-09-2005, 02:47 PM
EnJoY

Quote:

Originally Posted by burningrave101

Anyone else? I'm using 3120 at the moment as well but a tREF value of 3120 is 200Mhz MEMCLCK 15.6us and 1GB DIMM's aren't going to have a higher refresh rate then 7.8us. I'm surprised their not 3.9us considering how 256MB DIMMs are 15.6us and 512MB are 7.8us.

What about drive strength combinations? Do most of you just leave it to Auto or are you using normal or weak strengths?

Try, 166MHz 7.8us It's like 4676 in bios I think. Gets me 20mbps or so more bandwidth than 3120.
07-09-2005, 04:26 PM
nugzo

4672 :)
07-09-2005, 04:34 PM
nugzo

Timings

What timings should i tweak next for performance (besides the 3-3-3). Ive tried tref values of 3120, 3072, 1560, 0780, 4672 with no change in performance. 0016 tref was alot slower. Stability is great at the pictured settings. If this Memory was 100% stable at 280 i would drop multi to 10. 275 is max for stability. I havent done much tweaking at 275*10 (2750mhz), Do you think that could be faster than 255*11 (2806mhz)?

http://img150.imageshack.us/img150/3060/untitled1qf.jpg
07-09-2005, 04:48 PM
Vapor

x-2-2-y

as for 255x11 vs. 275x10....why don't you tell us? :D
07-09-2005, 04:59 PM
nugzo

Quote:

Originally Posted by Vapor

as for 255x11 vs. 275x10....why don't you tell us? :D

:p: cos thats alot of work :p:
07-09-2005, 05:10 PM
Vapor

LOL, I know...I don't want to do it either...

Anyway, I think it will actually vary by app, which sucks. Can you get 276 or 277 from your Ballistix? Maybe that'll make it a little clearer.
07-09-2005, 05:20 PM
nugzo

I might be able to get 276 or 277. i went from 275 to 280 and didnt try anything in between. i'm 280 stable on memtest but not super pi 32m :( I'll play with it tonight and see.
07-10-2005, 03:44 PM
EnJoY

New rule!

3. Super PI 32M stability screenshot is now a requirement for the database.

For those of you already on the database, just send me your super pi 32m screen within the next 10 days and I'll update it for you.

Show 100 post(s) from this thread on one page