do you tried CH5 on your A64 ??? it's the same on P4.
if you set max ram frequency at 166 mhz in bios you can't hit more than 200 mhz ram
With BH-5 desynchro works good , but my giga 939 hate them :(
Printable View
do you tried CH5 on your A64 ??? it's the same on P4.
if you set max ram frequency at 166 mhz in bios you can't hit more than 200 mhz ram
With BH-5 desynchro works good , but my giga 939 hate them :(
Hmmm, I tried some Kingston HyperX CH-5...
Yet another reason not to go S939 ... :D
I was under the impression that 939 loved CH-5, did 2-2-2-5 at clocks similar to BH-5. :shrug:
yes i said that already.
But , do you tried 166 mhz mode ?
and why do you said "Yet another reason not to go S939 " , i think it's a good thing , CH5 is very good, available, and cheap.
My setup is ready , i tried:
2200mhz 200X11 with 2/2/2/5 X800XT 575/575 : 24950 pts 2K1
2200mhz 220X10 2/2/2/5 with same vc freq : 25435 pts 2K1
2200mhz 244X9 2/2/2/5 ............................ : 25800 pts 2K1
i was wrong when i said there is just a little gain when increasing FSB on 939. :eek:
If i remember , my friend did 26600 with his 3200+ at 220X10 with same vc freq but under 2003 server. It will be hard to do better than Clawhammer 754. I must have about 100-150 Mhz more to perform as well.
1600mhz DC 200X8 2/2/2/5 520/560 : 20250 2k1
1924mhz DC 240X8 2/2/2/5 520/560 : 23346 2k1
2000mhz DC 250x8 2/2/2/5 520/560 : 24192 2k1
2400mhz DC 240x10 2/2/2/5 520/560 : 27002 2k1
2400mhz SC 240x10 2/2/2/5 520/560 : 26130 2k1
1600mhz DC 200X8 2/2/2/5 520/560 : 11655 2K3
1925mhz DC 240x8 2/2/2/5 520/560 : 12221 2k3
2000mhz DC 250x8 2/2/2/5 520/560 : 12319 2k3
2400mhz DC 240X10 2/2/2/5 520/560 : 12738 2k3
2400mhz SC 240X10 2/2/2/5 520/560 : 12508 2k3
In light of what misteroadster has shown us do I really need to bother? I don't have a chip in hand so I'd have to go buy one, and I don't have my OS configured that way. I've done enough testing to know that the ClawHammer will trounce those scores.
Is there anyone out there with a similarly configured (O/S, DX9, XT) ClawHammer who can run the benches exactly the way misteroadster did? That way I can concentrate on FX-53 benches today. I'll run them if need be but it would be cool if someone else could help us out.
Thanks!
hummm... yesterday you said you had a 2800+ , it's not cool , i'd prefered go out under the sunlight instead of bench for nothing :(
Moreover , i used 8X specially for you.
i'm not sure you can trounce my 2K3 scores.
3dm2k1 is a benchmark obsolete now, directX8 does not use the bandwidth as much as direct X9. 2K3 represent better the performances in the today games i think.
You didn't bench for nothing. I'll run the benches if nobody else will. You should never bench for others though. Always bench for yourself:D
of course but the frequency used are very poor :D
But we could learn the extra bandwith of DC is not so useless in fact.
you can do these benchs , remove one stick to your FX53 and you have a 3700+ :smileysex :D
then you give it the second , and we ll learn a lot about DC and L2 power .:toast:
Do this with the K8nsnxp to compare exactly.
That's a good idea.
Are you using the Asus or Gigabyte?
Oh, the Gigabyte! OK.
ready?? BENCH!
I updated my benches post with a compare Single/dual
You're right, CH5 is good, available, and cheap. But you said it doesn't work well with your S939 board, or am I missing something ?Quote:
Originally posted by misteroadster
yes i said that already.
But , do you tried 166 mhz mode ?
and why do you said "Yet another reason not to go S939 " , i think it's a good thing , CH5 is very good, available, and cheap.
So why did your SC vs. DC compare show the biggest gap in 2K1 ?Quote:
Originally posted by misteroadster
hummm... yesterday you said you had a 2800+ , it's not cool , i'd prefered go out under the sunlight instead of bench for nothing :(
Moreover , i used 8X specially for you.
i'm not sure you can trounce my 2K3 scores.
3dm2k1 is a benchmark obsolete now, directX8 does not use the bandwidth as much as direct X9. 2K3 represent better the performances in the today games i think.
I hardly think 2K1 is obsolete :D
humm, i said you CH5 don't like desynch , if you have an xtremecooling and a small coeff it's an handicap.Quote:
Originally posted by LilGator
You're right, CH5 is good, available, and cheap. But you said it doesn't work well with your S939 board, or am I missing something ?
But if you have an FX it's not.
I never had problem with CH5 and 939 , just with BH5 but it's probably my motherboard.
i think an increase of about 850 pts(=85mhz) at 2K1 is more less than 230 at 2K3 (=220mhz).Quote:
So why did your SC vs. DC compare show the biggest gap in 2K1 ?
As you can see beetwen 1925 (240X8) and 2400 (240X10) there is 475 mhz and about 500 pts at 3DM2K3 and about 4000 pts at 2K1.
When I issued the challenge it was with the idea that you use any settings and any Newcastle you want, and I'd use a 3200+ with similar cooling and see who can get the highest score. I didn't mean for you to run 20 benches. I don't really have time to do that. I'll run a few though. I can't get the memory clocks you have because my S939 board has no mods and my BH-5 and BH-6 require lots of volts to do anything.
Things aren't going so well. I can't get my A8V to run single channel. It doesn't want to post. Here are some DC scores. I'll have to run the single channel stuff on the S754 setup which will take some time to get going. Here are some benches for now. If you have time you could run these exact settings for a more accurate comparison.
All benches run on stock air cooled CPU and GPU, 72 deg F, closed-box, 6-2-2-2, 1T, no OS tweaks, XP SP1, DX9.0c, Cat 4.7, all virus checkers, spyware blockers, etc running. This is my 24/7 internet and gaming machine.
2100mhz DC 210X10 2/2/2/6 520/560 : 26488 2k1
2310mhz DC 210X11 2/2/2/6 520/560 : 28116 2k1
2100mhz DC 210X10 2/2/2/6 575/575 : 27180 2k1
2310mhz DC 210X11 2/2/2/6 575/575 : 28856 2k1
2310mhz DC 210X11 2/2/2/6 575/575 : 13940 2k3
All test run all at once, not 1 at a time, and no waiting between tests. I was going to run all the same on single channel but couldn't get the board to run. I'll do it on my S754, but like I said that will take a bit!
I'll update this with the 5 single channel runs later.;)
EDIT:
Ok, here are some S754 runs. I used a 3200+ (old C0 core) so i only had a 10 multiplier. To get 2310mhz I had to run 231mhz HTT and memory, so those scores are higher than they would be, but the 210X10 is identical. Also, the board for the S754 stuff is a Gigabyte so the chipset is very close to misteroadster's.
2100mhz S754 210X10 2/2/2/6 520/560 : 26131 2k1
2310mhz S754 231X10 2/2/2/6 520/560 : 28261 2k1
2100mhz S754 210X10 2/2/2/6 575/575 : 26693 2k1
2310mhz S754 231X10 2/2/2/6 575/575 : 28598 2k1
Interesting that the S754 actualy scored higher than the FX-53 with the 2310mhz (due to the higher buss) with the higher VC clocks. Higher memory speed wins out over DC bandwidth in terms of how it affects 3D performance. Also, it is VERY apparent that the 1MB cache vs 512k is in fact HUGE!!! Where have we heard that before I wonder? Wow, just as I'd predicted. I must have ESP. Or maybe I wasn't full of $hit when I said I'VE ALREADY TESTED THEM!!! :D
i don't understand why you don' use the gigabyte , here you use another chipset totally different , the via perform better but overclocking is poor with it.
but i will re-power on my system to bench at the same freq than you.
Of course FX53 scores are very impressive man , my god with this processor you would be far behind me on the orb !! :slobber:
2100mhz DC 210X10 2/2/2/6 520/560 : 24637 2k1 ; 1850 less than you
2310mhz DC 210X11 2/2/2/6 520/560 : 25786 2k1 ; 2330 less
2100mhz DC 210X10 2/2/2/6 575/575 : 24700 2k1 2480 less
2310mhz DC 210X11 2/2/2/6 575/575 : 26093 2k1 2800 less
2310mhz DC 210X11 2/2/2/6 575/575 : 13230 2k3
it's very strange , for me i have a poor gain between 520/560 and 575/575 : 400 pts and 63 pts and you have 700pts at each bench :confused:
I must have a problem with my vc; perhaps because i used a pro bios
Edit : My Vc is dead i think , i have a lot of artefacts.
i don't understand , there wasn't reason.
Even at 574/575? That's not good.
No, the reason for the difference is the cache. Cutting the cache in half robs the A64 of much of it's power.
The IBM PC clone architecture has gone largely unchanged for the past 25 years. The only major difference between what we had 25 years ago and what we have now is super fast CPUs due to two things primarily. 1) The process has gotten better and 2) large amounts of onchip cache used in creative ways. Of course you are going to weaken a CPU by cutting it's cache in half.
Getting ready to run single channel benches now.:D
me i'm ready to RMA this poor X800. :D
What happened?
i had suddenly full of artifacts, a kind of blue checkerwork in back
plane
Condensation? What cooling on card?
no condensartion
cooling likid at -5°
i don't understand , i'm nervous,i'm a lot more , i think i better to do the world walking , since I am small i think of it , perhaps nike is right.
i just have to do it :D
i think it's cheaper than overclocking and a lot more a trip :rolleyes: per haps i drunk too much this night , but you could understood. my poor X800 is dead
i want a real life with adventure.
OC close you in your home. a day your setup died and you don't understand.
i just understand that , i'm so sad. :(
Sorry to hear about what happened to your card my friend. Good luck with the rma. I felt the same way when I cracked the core on my 9800XT.
thanks phil , I am stuffed a little less, i hope to have another one tomorrow ;)
misteroadster, i just had a problem like yours in my X800, the condensation on the back of the card have created a green stuff.
I´ve brush it with tooth brush, cleen it very well and voilá it works :D
thx master , but , i don't have green stuff on the back , she's dead , it's sure.