Got it!Quote:
Originally Posted by |RickY|
Thanks. Benchmark time :)
Printable View
Got it!Quote:
Originally Posted by |RickY|
Thanks. Benchmark time :)
Quote:
Originally Posted by OPPAINTER
my math puts your ram at 255fsb but it doesnt matter with an amazing score like that anyway. hes probably tried every multi and this was the best compromise. Are you done till san diago fx57?
"Thanks. Benchmark time"
does this include 3dmark 2001? We think you can break 40k with your skills :slobber: :D :p:
wtf? i always thought that to continue to stay on top you continously improve tiny bits and pieces. so why would he be done till fx57? wasn't it just a few weeks ago that the fx53 did score big time? according to your statement they should've been done around a few months back. silly comment, don't you think?Quote:
Originally Posted by Geforce4ti4200
macci, now i want to see the cascades of doom unleashed :D
easy to know : 260X13.5/14=250.7mhz.Quote:
Originally Posted by OPPAINTER
;)
it includes 2001 only :DQuote:
does this include 3dmark 2001?
just testing different settings w/ aircooled winchester rig for starters.
useing 3500+ @ 230x10, 1:1, 2-2-2-5-7-16 and stock X800Pro w/ 4.10 driver I've got these reference scores so far:
WinXP SP2 => 25248
Win2k SP4 => 25823
Win2k SP4 w/ a64 tweaks => 25918
Gotta see if I can find the next tweak level ;)
Still the same clocks (230x10, x800pro stock):Quote:
Originally Posted by macci
Win2k SP4 + "ricky special" + A64tweaker => 26388
I think I found the next level :D Thanks to Ricky for pointing me to the right direction - still took me 4 win2k installations and several hours to finally get there ;)
Full test data :D
http://www.akiba-pc.com/939/testdata.jpg
Device Manager is your Friend :D
That looks very interesting.. it is a nice boost the rickY special gives. But is SP2 not better than SP4.. You said that your self .. ! when are you going to put some nice cooling on the cpu soon?
Wow! 10 fps in car hi just from using properly tweaked OS....impressive!!
Quote:
Originally Posted by Geforce4ti4200
lol. :DQuote:
Originally Posted by misteroadster
i have a feeling theres no chance that anyone else will get to know this "ricky tweak"?? :p:
i wouldnt tell anyone.... go macci go.
It would be very cool of |RickY|, Macci and Pedro Rocha to share the special tweak, then they could get some more competition on the top spots. But if they are competition men, that would do everything to be in the top and doesn't want any kind of competition then they keep it for them selfs.. In details would be fine :D
I agree:DQuote:
Originally Posted by JacobHansen
OPP
I knew you would :)Quote:
Originally Posted by OPPAINTER
Brian and me are the kind of people that likes the competition on the Top, and would like to share the mods, cooling tricks, tweaks etc. But only to get the trill of trying to be number one. And i do se the point in following a guy that is f.eks 1000~ ahead of me, and we are at same clocks.. But it is very hard to be able to catch up with you guys IF you dont share the mods. :D But we'll indeed try as hard as we can to beat you :banana4: Hope it makes sense :p:
Macci - Nice job investigateing ;)
our best scores compared to opp are explained, period
we are not cheaters
our best car/ lobby isn't cheat, is performance boost based on a better tweaked system
remember how we learned to share in kindergarden ;)
Macci knows the ricky tweak none of us do. No one is calling Ricky a cheater. I dont mind if me nor anyone else gets to know those tweaks, ill never come close to the top anyway but it would be nice if Macci or Ricky shares those with OPP ;) just friendly competition ;)
Nobody is calling you cheater..Quote:
Originally Posted by |RickY|
I allready know that ricky - Thats why i was asking macci and you to tell them to us :D
But we are gonna locate them ourself it seems..
Ricky U should have shut up dont have to prove anything to anyone
skatan my friend, nice to see you :)
:toast:
And your going to feel a lot better in the end that way :)Quote:
But we are gonna locate them ourself it seems..
The info I've given in this thread is enough to figure it out if you know what your doing :D
Actually its more than Ricky gave me ;)
this probably best... o/s tweaking is certainly benchmarking, and I see no reason why it MUST be made public. We've been given a few clues, let's find it ourselves!!! If we can ;)Quote:
Originally Posted by macci
C
i think the only clues we've been given is device manager, and win2k sp4...
maybe the test order scribbled on the bottom of the paper, but i doubt thats anything to do with the tweak...
obviously the hint starts here:
obviously... :stick:Quote:
Originally Posted by pik-ard v1.1
The Standard PC /ACPI thing?
I know their is a point gain when you change to uniprocessor PC
Can some of you 3dm gurus help me with this problem? http://xtremesystems.org/forums/showthread.php?t=44754
Quote:
Originally Posted by µnrealneo²
We do not talk about XP. It's W2K.
It has! But maybe with a different effect on 3DMark performance?Quote:
Originally Posted by µnrealneo²
well if this tweak becomes public, me and everyone will use that. In my case ill get like 200 marks boost :) others will gain 300 or 400
ExactlyQuote:
Originally Posted by SKATAN
yah when he said device manager i knew theres really only one thing there that could have a big effect on performance ;)
I think I know too, gonna try and see. If no go, itll be public knowlege soon
I hope those who know what it is that's being done keep there mouth shut.Quote:
Originally Posted by P4rthen0n
the question is, are o/s tweaks fair game? Obvious answer is YES. But WHAT about a custom designed o/s? What if a MS engineer made a skeleton win o/s program just for 3D benching full of optimizations? What if it gave an extra 1000 points? Would it be fair? Where's the line? How 'bout wireframe 3D mods? Naughty, right?
What's the RULE for "acceptability"?
Software and Hardware available to the public?
Tweaks and Massaging no limit?
Hmmmmmm...
Seems to me that the old "set priority to real time" is acceptable and is known to everyone. Obviously the |RICKY| mod is OK, too.
While the ricky mod is a bit more in-depth there are definately limits.. otherwise we would all be running super_pi on *nix with WINE maybe? :D
Nice find btw ricky ;)
From the view of being nr.1 I truly understand Mr.Rocha, Ricky and Macci :toast:
Just thought Xtremesystems was all about sharing info and helping each other getting better :)
Still at 230x10 but faster card (X800XT PE):Quote:
winchester@2300MHz, 230x10, x800pro:
WinXP SP2 => 25248
Win2k SP4 => 25823
Win2k SP4 w/ a64 tweaks => 25918
Win2k SP4 + "ricky special" + A64tweaker => 26388
Win2k SP4, all-out tweaks, stock X800XT PE => 28611
Win2k SP4, tweaks, X800XT @ 669/630 => 29603
I'm fairly sure that a 512k cache CPU at 2300MHz w/ 230MHz ramspeed cant score much higher than that :D
One thing is funny thou: I'm getting silly low Drago low score?? is that cos of 512k cache?
comparison w/ my 1MB Clawhammer:
http://www.akiba-pc.com/939/claw_vs_winchester.PNG
512k is at 2300MHz (230x10), 1024k is at 2394 (266x9)
660/630 vs. 567/612
Same thing happened to me :confused:Quote:
Originally Posted by macci
Havent worked out what kills drago low yet, maybe it was something wrong I did during the win2k install. At same GPU clocks my 3GHz S754 is beating my 3.1GHz FX53 ... weird.
** sigh ** another problem to work out.
Hmm ok so its not a cache problem then? how much difference did you see between s754 and your fx?
tried WinXP too and it was even slower :D
I cant find an exact example in my notes but this is the best example of the drop:
3400+ @ 2970 (11x270), GPU 610/610 -> DL 674.3 FPS
FX53 @ 3127 (12x260), GPU 651/642 -> DL 674.2 FPS :(
same OS tweaks, same AGP settings, same mem timings (full A64 tweaks), same test order, and both CAT4.9.
The FX53 should flog the S754 score :confused:
Only OS difference I can find is that I used 3.13 NVidia drivers on the S754 and 5.11 + M$ IDE driver on the FX53. Also used LDT2.5x on the S754/shuttle and LDT4x on the FX53/giga.
I was thinking that it could be an AGP latency problem or a chipset driver issue, gotta test more to find out.
Drago Low is a VERY "Memory depenting" bench.........Try to run it with HIGHER Ram frequency and you'll "find" your frames/sec...... ;)Quote:
Originally Posted by macci
EDIT1 : EVEN 1 or 2 MHz more on Ram frequency makes difference.......
EDIT2 : I also found out that running the VGA at it's "edge" frequency for the DL , makes things worse.......Better lower a "bit" the VGA's frequency when running the DL + More rig's Ram frequency..... ;)
I think macci knows ;)
Gomer - it is, but you cant reveal everything! That'd take the fun and adventure out of learning all this crap ;)
Its 10000x more rewarding to figure out the trick than having it told to you.
If you know it, cool, keep it. If everyone shared all the damn time, wed all have the same scores. How much fun would that be? :stick:
Yes macci might know but there might be some others out there who don't........So I don't think it's wrong to share our experiences!......Is it?...Quote:
Originally Posted by STEvil
So you will share the spec's/design for the Maximizer?;)
In my opinion (ok its worth nothing :stick: )
All us small fish need a helping hand from the big fish :D
Give us half the tweak lol its a start :D i just killed win2k 3 times in a row lol not good :stick:
Spec's are shared and they should be.....BUT the design cirquit it's NOT a tweak or a mod or something like that....so it just can't be shared.......I've shared a LOT of other things though...... ;)Quote:
Originally Posted by G H Z
I dont think its necessary for anyone to divulge the tweak unless they want too. The pressure we are putting on ricky/pedro etc etc is just not fair. They worked it out the hard way, and not by having someone tell them how to do it.
Anyway, macci hinted enough for us to work it out.
I was curious about this, but I only have WinXP. I did do one little device change, the one I thought was obvious, and I don't know if this is it. I got ~23.9k after device change and ~23.5 before. Changed back and still get ~23.5k. Some scores was raised, while some was exactly the same +/- 0.? ...
I doubt this is it, as I have WinXP Home and not W2k. I'll share what I did, if anyone wants it, but this it probably not it.
Test things. Top scores reward top work.
I've never had a machine capable of the numbers that macci, DJ, or OPP put down but in the class I compete in I've had unrivaled scores that spanned 6 months.
Then a pack of wolves :banana::banana::banana::banana::banana: and cry and moan over OS tweaks I use and wont share. But ya know what? It took me time to figure them out. It ought to take them time as well.
word :)
dont get pissed ;)
u are right
Totally agree STEvil, but still thinking sharing info like Hipro5 just did and other tweaks and voltmods is great. I and others (including some of the "big" ones) aprriciate when people are sharing mods from you, Hell-Fire, KYOSEN etc....Well they could just find out about it themselves :) (kinda of bad Danish humour :) )Quote:
Originally Posted by STEvil
Not just for 3DMark but in generel for optimizing forum users 24/7 settings etc.
Well just an opinion, not for starting an diskussion (or what its called in english :) )
btw. gratz on the fine tweaks Macci and the aswsome results Opp.
No, don't think so.Quote:
Originally Posted by Hallowed
Maybe not but it takes the fun out learning.Quote:
Originally Posted by I²K
But it would be a nice competition :-)Quote:
Originally Posted by iceman2g
The more I surf thru the ORB the more I see 512k chips getting poor drago low scores and 1024k chips getting good drago low scores. I guess its just the cache making the difference in that particular test. Difference appears to be quite huge thou - around 50 FPS at default speeds!Quote:
One thing is funny thou: I'm getting silly low Drago low score?? is that cos of 512k cache?
But I need to test w/ 1M s939 chip to confirm this :D
I think you're right about DL & cache Macci; I was looking at some scores the other day and noticed the exact same thing; the 512k loses massively compared to the 1mb. These are the scores I was looking at: Stevehat1's 32K with s939 512K and his 32K with s754 1MB. As you can see, his X800 XT is clocked very similar on both runs (4fps diff), so it's the system that determines those scores. The s939 512K Winchester @ 2.71 beats the s754 Clawhammer @ 2.64 in all tests, except in Drago High where they're tied and DL where the 512K gets it's ass handed to it by the Claw. :) Should help clear up the fact that DL really likes cache :2cents:Quote:
Originally Posted by macci
Or DL likes dual channel?Quote:
Originally Posted by twilius_basic
Quote:
Or DL likes dual channel?
So no its not a DC vs. SC issue since the DC 512k chip is some 50FPS slower clock to clock compared to SC 1MB chip.Quote:
and DL where the 512K (dualchannel) gets it's ass handed to it by the Claw (1MB, Single Channel).
Epox 8KDA3+ Vcore/Vdimm modded //
AMD 64 3200+ XMPW 1MB @ 2.7Ghz //
OCZ PC3700EB @245Mhz 3-2-2-5 //
Club3D X800XT PE //
nVentiv Mach II Cooled
30450 3DMarks http://service.futuremark.com/compare?2k1=8049314
DFI Lanparty UT nF3 250GB 11x264 // 1024Mb OCZ 3700EB DDR528 SPD timings // Club3D Ati Radeon X800XT Pe Cooled by nVentiv Prometia Mach II
28452 3DMarks http://service.futuremark.com/compare?2k1=8167236
The NewCastle was meant to replace my 1Mb cpu, but now I've ordered a 3400+ 1Mb to replace this NewCastle. My old 2.7Ghz CH was just way faster. I used to play UT2004 on it with the CH @2.5, with the NC I played UT2004 @2.9Ghz. And there were some NASTY fps drops with the NewCastle, down to like 15fps when I was playing RPG Invasion and the Skaarj showed up (fps with the ClawHammer never dropped below 60). Especially wave 5 where plasma bolts are going all over the place :cool:
The NewCastle has 200Mhz on the Clawhammer and more memory bandwidth and it loses with around 3k points.
My opinion, NC's are like Durons.
Just goes to show that 90nM FX, 1M will be a real ROCKET :DQuote:
Originally Posted by macci
how can the newcastle be slow and the claw fast if theres 200-300 MHz difference? a 2.6GHz newcastle should run games as well as a 2.3GHz claw. Yes drag low sure likes teh cache, but so do all other tests except nature and car high doesnt care too much. car low is the 2nd most cache intensive test. Yes the 90nm fx57 should be a rocket, dubbed "san diago" those monsters will start at 2.8GHz and probably used strained silcon and overclock to around 3.1GHz with high end air cooling
Ace, its San Diego.
what I really want to figure out is how to enable the extra 512k on my 2800+ clawhammer :D
http://service.futuremark.com/compare?2k1=8219965
OMG OPP breaks 42k marks!!!!!!!!!!!! 42330!!!!!!!!! 800+ drag low, 500+ lobby low!!!!!!! :slobber: Nice 340 nature, 186 car high too! such a shame win2k doesnt work on this fx53 or it would be like 42.5k easy. I never imagined 42k would be broken for 6 months, but that was because I did not think the fx55 would overclock better than the fx53! ~3.6GHz is just darn amazing!!!!!!!! You got me very curious what youd get on your 9800xt now ;) 34k is my guess with that new cpu and the latest 4.10 cats. :p: wanna find out? ;) :D
your 9800xt score:
http://service.futuremark.com/compare?2k1=7382155
does the 4.10 cats boost nature any? They boost the fps of all other tests plus you now got an extra 500MHz cpu power. should easily account for another 3.7k for 34k ;)
Stunning. Look at the cpu speed. - I am outta words
wow, I am speechless
congrats OPP, that is truly amazing :toast:
Dayamn.
Way to go OPP :toast: :slobber:
interesting to see you used > Microsoft Windows Server 2003 think its time to install it and see :D
Quote:
Originally Posted by stestan
dont bother, OPP tried using win2k but his fx55 for some strange reason isnt stable under win2k when superchilled. Id have to see if others are able to run their fx55 superchilled under win2k, if not then something in the fx55 is the curlpit.
Good Job OPP
42k :slobber: :slobber: :slobber:
Congrats
Héhé , 2K3 server is a good OS for 3D benchmarks ;) congratz again OPP :toast:
Opp tearing up the 3D, lets see a high end screen shot from that rig. Might be a world record. If you can add Superpi to your bench drive that would mucho appriciated by all.
Sandra memory bandwidth shot would also be nice too.
Awesome job once again
Looks like OPP got the settings figured out :D
Simply amazing! :toast:
Congratz OPP, that is amazing
:toast: :toast:
The portuguese have to work hard now.... :) :)
Great score :toast:
congratulations :toast:
:toast:
Nice work Opp.
What was the high fps in drag low you saw ?
Regards
Andy
congrats OPP :banana: :toast: :toast: :toast: I'm glad for u :p:Quote:
Originally Posted by misteroadster
Win 2k3 is the best for all the benchs except the PI 32
Opp, u think u can beat 3821MHz in screenshot with ur FX?
Unstopppable~~
Ownage Mr Opp! :toast:
42k is beyond sick..! in a good way of course :DQuote:
Originally Posted by Geforce4ti4200
I was surprised it wasn't higher.Quote:
Originally Posted by zakelwe
I think I saw about 925, can't remember:D
OPP
Give me a download link to this Superpi benchmark, I'll see what it does.Quote:
Originally Posted by FUGGER
OPP
Thanks,Quote:
Originally Posted by FUGGER
I'll play with it today.
BTW:
I need that M-Wave order#:)
OPP
I can't seem to get to the SuperPI,
seems the link they provided changed locations.
OPP
http://pifast.hexus.net/hexus_pifast.zip <-- Thats pifast. Popup download from the pifast list.
what a bost opp, han? :D
http://users.pandora.be/cranox/Progs/Superpi/ -> for superpi ,this link works
Gl luck opp ;) ( and don't vote 4 bush today :D)
OPP, the World Record for SuperPi 1M is 23sec with FX@3557MHz (13.5x263). I think you can pull a 3.6G out in that bench - 14x258 :D
here is the ranking for the SuperPi 8M test which IMO is a better challenge than the 1M test.
Those wacky Japanese, imagine congratulating some called New Beetle or Joy .
" :toast: Well Done New Beetle, now that's what I call a bugged score ;) "
No, I'd rather stick to our sensible ;) nicknames
Regards
Zakelwe
And by the way, if the Japanese refuse to come out and play in 3dmark it is only fair that our tops guys from USA and Europe go over there and take them out in SuperPi :)
Sort of like a Ryder cup for overclocking Japan v Europe/USA. Cascades v LN2
Regards
Andy
try PiFast instead... pretty please with sugar on top :DQuote:
Originally Posted by OPPAINTER
Yea Ricky, thats some tweak you've come up with:DQuote:
Originally Posted by |RickY|
OPP
I know that noone is spilling the beans on this tweak, but I'm just curious, would it still work on Win2K + Barton? Instead of A64....Quote:
Originally Posted by OPPAINTER
Wouldn't it be unfair to memesama to call the other 23s-guys record holders as he was the one who scored it first? :)Quote:
Originally Posted by macci
Quote:
Originally Posted by OPPAINTER
This gained ya 700 marks if im right? The gains with win2k are less since win2k is already more streamlined if I recall right :D by the way, still got that 9800xt? would be interesting to see what she can do in your rig ;)
http://www.alanpotter-publicart.com/...ing%20john.jpg
42K - OPPStyle :D
Congratulations my friend, you are the 3DBench King, and we are mere prince regents that can take your throne only in the brief moments when you are busy to buy more powerhouse to your heavy cavalry :D
Here is the compare with my poor FX-53, that rest now on my Promie until we get some gas to our light horse cavalry :)
http://home.kqnet.pt/wo034892/MeAgainstOpp42k.jpg
I'm sure you will get some more gas to your stallion soon Pedro.
Thanks for you and Ricky sharing your tweak in a fun and puzzling way via macci, it's appreciated by the community :toast: Hopefully you will be paid back in kind, or more so, due to your generosity.
Good luck with the gas, good luck with the new FX-55 I am sure you have up your sleeve ! :D
Regards
Andy