updated, i just picked up some DDR400 ram from local computer shop today.:)
1gb to be exact. more benchies to come!:toast:
Printable View
updated, i just picked up some DDR400 ram from local computer shop today.:)
1gb to be exact. more benchies to come!:toast:
woot finally broke 17k
:toast:
http://service.futuremark.com/compare?2k1=7979357
DDR400 and PAT gave huge boost in bandwidth.
:toast:
Which would you rather have, 5810 with about 600MB of overhead or 6136MB (lasted test ican remember at 250 1:1) and almost 2GB of overhead. Now RAM might have gotten better since we compared them last year but then, that much overhead cause more of hit, most of the time. Overhead = Latency. Yet, as was stated, there are times when more bandwidth is better than less that starts faster.Quote:
Originally posted by Soulburner
Dropping to 5:4 after being at 1:1 can cost you as much as a 250mhz clock speed drop in 3D apps (shown by macci...notice how he "ALWAYS" runs 1:1? There is a reason).
Again, a hit on i865 is not a factor on i875=P PAT runs on i875 all the time unless you got crappy RAM. The Ratio is there for the memory, not the FSB that will produced the same amount of bandwidth no matter where the RAM is set.
Other have seen this happen as well. I use to have 3 links, but
here's one.
http://www.overclockers.com/articles758/index.asp
If you guys got this to work better then great, hell, fill me in on it?
I'm all for it and might jump on 1GB of PC-4400!
http://jupiter.walagata.com/w/donnie27/3464MHz.JPG
Donnie27
You're welcomed.Quote:
Originally posted by Kanavit
good read, thx Donnie27. very useful info there. I may be getting DDR400 pretty soon. I'll keep everyone posted.
It's what I've seen, but hey, I standby the Ratio info. If the other info has changed, I'm happy and don't care about being wrong. When we tried this, the fastest RAM out was PC-3700 and the newest PC-4000 in those days were a rip off, not like the good stuff they're selling now.
Donnie27
why not overclock? Prescotts run hot, but they can also take a lot of heat.
DDR 400 price seems to be rising. they are pretty expensive.
well i have mine and i aint letting it go.
Time for ddr2 to drop in price.
Wonder if they are stopping making ddr so the price is going up.
could be. lets see how long i hang on to these DDR400 dimms. my DDR333 is collecting dust now. don't feel like selling em. mabye they will go in my other system.
The fact of the matter is macci could not beat his 3DMark01 scores at 4200-4300mhz in 5:4 mode that he set at 4000mhz in 1:1 mode.Quote:
Originally posted by Donnie27
Which would you rather have, 5810 with about 600MB of overhead or 6136MB (lasted test ican remember at 250 1:1) and almost 2GB of overhead. Now RAM might have gotten better since we compared them last year but then, that much overhead cause more of hit, most of the time. Overhead = Latency. Yet, as was stated, there are times when more bandwidth is better than less that starts faster.
Again, a hit on i865 is not a factor on i875=P PAT runs on i875 all the time unless you got crappy RAM. The Ratio is there for the memory, not the FSB that will produced the same amount of bandwidth no matter where the RAM is set.
Other have seen this happen as well. I use to have 3 links, but
here's one.
http://www.overclockers.com/articles758/index.asp
If you guys got this to work better then great, hell, fill me in on it?
I'm all for it and might jump on 1GB of PC-4400!
http://jupiter.walagata.com/w/donnie27/3464MHz.JPG
Donnie27
1:1 IS THE KING. But NOT with 3-4-4-8 timings.
That is a real world benchmark. Sandra is going to say the opposite, leading you to believe these silly things...your PC4400 will likely be slower than any BH-5 or OCZ EB ram no matter how high you can get it.
First for the memory, I don't need to set the RAM to CAS3 or 2.5 8-4-4. All it has to do for me is 270-215 and at least CAS2 6-2-2, anything else is pudding! Seeing a 2.8C run at this setting was and is enough for me.Quote:
Originally posted by Soulburner
The fact of the matter is macci could not beat his 3DMark01 scores at 4200-4300mhz in 5:4 mode that he set at 4000mhz in 1:1 mode.
1:1 IS THE KING. But NOT with 3-4-4-8 timings.
That is a real world benchmark. Sandra is going to say the opposite, leading you to believe these silly things...your PC4400 will likely be slower than any BH-5 or OCZ EB ram no matter how high you can get it.
Ace's Hardware showed what most of us who think 3DMarks is more of a Video Card Benchmark and a good system stress test, not much else. 2001 isn't all that and 2003 stinks even worse.
The difference with Sandra is that its scripts are optimized and even some of the tests show unoptimized/optimized results. Many of the tests on 3DMarks don't relate to current software, namely Games. The games that make up the bench mark is not the same as running some of the stuff from Bapco or etc..
From Ace's;
http://www.aceshardware.com/read.jsp?id=60000242
After testing both 3DMark03 and a variety of real-world games based on DirectX 7, 8.1, and 9, we suspect that 3DMark03 is really more of a video card benchmark than anything else. It seems that some of the game tests, especially Mother Nature, measure video card performance almost in isolation from the rest of the system.
As far as Sandra goes, lets see how much Hyperthreading and SSE2 or 3 Doom3 has? If there were Icoming Forces (optimized for SSE2 from the ground up) test results, we'd see just how close Sandra is to telling the turth. It's not SysSoft Sandra Fault Games aren't optimized worth a fart! Bad that sucks on Intel's part for making it so dependant on SIMDs.
The Newest Adobe Premiere Pro with Hyperthreading and SEE2 shows very similar results as Sandra.
Donnie27
I was talking about 3DMark2001...which is a very system speed sensitive test.
OK, I still say it (2001) is only good for testing like systems more than anything else.
Donnie27
I don't agree...whicher system pushes the higher FPS there is faster IMO.
And besides I was talking about like systems. 18x250 1:1 is faster than 320x14 5:4.
Nope, can't argue there, I've never been over 4GHz and can't say which one would be the fastest.
I just don't think even 3DMarks SE is a very good measure of real Game performance, no matter what system it is installed on. Please Run Game/Demo Benchmarks like Commanche4, X2, Gun Metal, Aquamark and etc.., these would be of more valueble than something none-real world like 3DMarks 01 or 02. I hope Doom3 comes with a neat benchmarking tool/feature.
Donnie27
Fugger, how far does that DDR2 clock with those timings?
Seems to be clock to clock pretty much just as fast as Bh5 at 200 1:1
http://www.akiba-pc.com/e107_images/...unbuf_1024.gif
i dont agree with that. i use pc4000 a-data and also have a pair of pc4200 and at 280 1:1 my memory scores and 3d scores are just fine. i also have a pair of crosair pc3200 bh-5 and even at the 2,2,2 settings i can stil out perform them with the pc4000 series. now i can only use 2.8v on the memory so that probly isnt fair to the bh-5 cause they like the higher voltages. anyways i just dont agree thats all. i may be wrong and if i am thats cool but i just dont see it...Quote:
Originally posted by Soulburner
The fact of the matter is macci could not beat his 3DMark01 scores at 4200-4300mhz in 5:4 mode that he set at 4000mhz in 1:1 mode.
1:1 IS THE KING. But NOT with 3-4-4-8 timings.
That is a real world benchmark. Sandra is going to say the opposite, leading you to believe these silly things...your PC4400 will likely be slower than any BH-5 or OCZ EB ram no matter how high you can get it.
peace perc,
Fuggers got 4 DDR2 dimms in that mobo. which is really interesting configuration. those sandra scores aren't bad considering the timings.
i thought 2 banks supported DDR , and other 2 supports DDR2 for i9xx chipset mobos?
I just know the top guys test all sorts of configurations and settings to find whats the fastest. Logically if it wasn't the best they wouldn't be using it.
What I stated about the 5:4 vs 1:1, he said himself...so its kinda hard to argue with macci.
:)
i925 doesn't support DDR at all, only DDR2Quote:
Originally posted by Kanavit
Fuggers got 4 DDR2 dimms in that mobo. which is really interesting configuration. those sandra scores aren't bad considering the timings.
i thought 2 banks supported DDR , and other 2 supports DDR2 for i9xx chipset mobos?
i915 supports both
thx Snowwie! :toast:Quote:
Originally posted by snowwie
i925 doesn't support DDR at all, only DDR2
i915 supports both
Macci, about 206FSB with tightest.
my 3.0E
SL79L
L410B186
http://membres.lycos.fr/cpulloverclock79/3.0e34s.jpg
not bad.