http://valid.canardpc.com/show_oc.php?id=471056
There you go.
Printable View
http://valid.canardpc.com/show_oc.php?id=471056
There you go.
Could it provide any performance boost when high HTT clocks are used like with high FSB on Intel platforms?
I think HTT can reach high clock is advantaged for overclockers who are not using Black Edition CPU :)Quote:
Originally Posted by Stuen4y
Hossa, :shocked: :cool:Quote:
Originally Posted by Macadamia
The htt link multiplier is at 1x though????
Is that even possible?
and why would they lower it that much?
Was it needed to make the high htt work?
If it'll work that high normally we'll see some really good overclocking on the 920 and other non black edition cores.
I can't wait to see how the dual cores will clock they should be able to clock up to wolfdale levels from how things are turning out so far.
For me is quite interesting fo future PII920 owners!
This COULD be the CPUZ bug as well. I don't hope so, but it's definately a possibility (at least until we see benches at 300+ MHz).:)
Makes me happy , I have that same Mobo :D
Heres a screen fom coolice
http://www.breakthelimit.net/coolice.../oc/maxHTT.PNG
http://www.breakthelimit.net/coolice.../maxhtlink.PNG
original thread here
http://forum.breakthelimit.net/index...&st=0#entry780
Lowest possible HT multi seems to be indeed 1x. I have not yet seen sub 5x.
This is the responsible register.
http://www.abload.de/img/htzbv5.jpg
Will try to modify this register with baredit tomorrow to see if it's possible to use lowert than 5x on the asrock and gbt boards whom only allow 5x min via bios. With 5x I'm hitting a wall at 280MHz and i'd need abit more to beat 9500 wr. :rolleyes:
wow so is there anyone else that has a phenom II or is it just coolice? he keeps pumping all this stuff out.
Any idea if this is expected to be common? I.E. can phenoms clock with HT better now? If so then I may not wait as long for an AM3 BE to come around.
since coolice is the only when i have seen do it im not sure if this is a simple thing to do. deneb is two weeks away and we should know soon.
Nice man... ;) The only thing really high HTRef clocks prove though is the ability for locked clock CPU's (such as the 920) to reach higer CPU frequencies...
As impressive as that is, my experience with Phenoms have shown that keeping the HTRef around 200 and using Multis for higher CPU/IMC/Mem clocks have shown better results....
Hopefully things will change with new Mobos, but from my experience high HTref clocks get a little squirley and cause wierd results...
In the end though 350 on the HtRef (benchable) is pretty damn impressive!! :D
I'd still rather have a 940 @ 210x20x12 at reasonable V's...
Only time will tell, but I don't see Xtreme HtRef clocks being the future of PhII...
Damn... I need a 940BE RIGHT NOW!! :yepp:
Any ideas what was the cooling for this 5GHz+ benches? It should be cascade or phase change for 1.7V+...Quote:
Originally Posted by Titan7171
P.S. It's great we have any opportunities to use some HTT in order to keep our memory frequencies at the edge of the chips' max.
P.P.S. Silly me it's written the tests were on L2N.
very good, so "non-black" series users will have a good oc to.
My previous MSI board had HT multis all way down to 1x in BIOS. At 1x it had problems to POST with X2 6000+ but 2x was working fine :)Quote:
Originally Posted by justapost
恭喜恭喜， 皆大欢喜 :D
Very nice scores, and thanks for being considerate to the 920 guys. :)
Damn now i am gonna get myself a 920 for sure!
Perfect !!
Great :D
Hi Toppc, can you tell me what HSF you used for that 4GHz system :)
Hmmm, DDR 1200 running 1:1 seems very doable:D Very nice results!:cool:
IN MSI 9600GSO 512MB 256bit
intel i7 920
http://www.coolaler.com.tw/toppc/96GSO512/06.JPG
amd x4 940
http://www.coolaler.com.tw/toppc/X4940/96GSO.JPG
I'd like to see a High HTT shot under DI ;) etc for 920 users looking for high clocks.
^^^^^ @Toppc
You forgot to cover SM2/SM3/CPU score on that phenom run :P: , i can tell you that for 4269 / 3943 / 5682 = 10711 ... using 3D mark clac i should see a 1 at the end not a zero or it is a rounding error , i wont call it a fake coz it might be just 10710 ...
Hello thx for the info :), i did see something dont forget to change your NB clocks :) , 1800 is very slow use 2400mhz+ if you can you will get more score in 3dmarkQuote:
Originally Posted by Toppc
the one your talking about is a i920 intel run with 9600nvidia gpu.Quote:
Originally Posted by kemo
so no reason to cover those right?
10k @ 4.2ghz hmmmm interesting.
so like i and and many others allready know. its about total platform bandwidth. allways has allways will be. matched parts just work better.
its funny. up untill nehalem. 9770 and lower cpus were the be all of cpu's. now they all talk as if anything other then i940 or i965 are junk . LMAO
i think this is what seperates those that know and those that just are google, forum read it, and its true junkies.
toppc im a little confused. are you saying that a phenom II where only the cpu multi is changed will give equal gaming performance as a higher oced i7 even with the same gpus? the i7 got a higher cpu score but the phenom II managed to get a higher gpu score with the aid of a phenom II? and in the end the overall 3d mark scores were just about the same. http://www.breakthelimit.net/coolice...3d06-17754.PNG i see coolice managed to get a slightly higher cpu score with a phenom II at 4ghz.
got anything to say or you just letting us figure it out? :D
Without sounding like the pessimist, I would agree with some of the questions raised in this argument. Ever since the introduction of the AMD demo a few weeks back we have seen a number of "apparent" benchmarks, Super Pi claims and a variety of other(so called) leaks. But the consistency of the information given has been so sporatic, that its very difficult to ever make any solid conclusions.Quote:
Originally Posted by roofsniper
All the signs seem to point at fraudulent posts and claims.
It just makes the entire thing that much harder to swallow now.
Just who is telling the truth at this point? :cool:
yes roof becuase you can up all the multis seperately no reason fot high core clock. cpu,nb,ht,ram are still going to run at there rated or overclocked speed. the bandwidth will still be there. so gone are the days of high core clock being needed.
so up up the multi till you wont boot check ram nb and htt. then slowly up the volts and htt. with a unlocked cpu that is. sur eyou allready know this but just for general info for those that dont know. also for those that dont know keep ht below or at same speed of the nb. :) happy clocking to all.
yes there were many benchmarks at first and many turned out to be fake. but as we are nearing release we are seeing constancy and some benchmarks match with others making us believe those are the true results. as far as this goes its been done in the past that a cpu with a lower cpu score can actually manage to get higher gpu scores. if this is the case it might be one of those things where even tho we have seen benchmarks where i7 beats out phenom II phenom II might end up being a better gaming cpu. notice what i just said is speculation and im not basing it on any facts but the 3d mark result is interesting nonetheless.Quote:
Originally Posted by Smalltimer
If the Phenom II price is right, I would have no doubts it could quickly become the ideal gaming/desktop solution. - I won't pretend to know what AMD is thinking(I don't follow roadmaps), but it would make sense for them to pursue this position.Quote:
Originally Posted by roofsniper
I'm still waiting for some hardcore high HTT/Overclocks to gather what the Phenom II can do in rendering. As I mentioned before, I would have no problems grabbing a pair of PII's in a mini-cluster if it falls in the same cost area as aCore i7. - I'm almost certain 8 Phenom II cores would outperform a Core i7(OC'd or not).
In fact, last night I dreamed of running a pair of Phenom II's @ 4.5Ghz under water under my desk. :p:
can you reduce the clocks on Intel Core i7 920 CPU to 3.8GHzQuote:
Originally Posted by Toppc
you sure you are running the GFX card at the same clocks eek
4.2GHz nehalem - 10814
3.8GHz deneb - 10711
Very impressive:shocked:
Although, then again FM benchies it ain't everything, I think that's a damn nice result:up:
hmm gpu score is higher but cpu score is a lot lower. There are ~100 points less on SM2.0/3.0 scores. Wonder where that comes from.
didnīt we see the same when b3 phenoms came out?Quote:
Originally Posted by Hornet331
b3īs clocked at the same speed as core2qīs got alot lower cpu scores in 06 but gpu scores were higher. so in the end the 3dmarks were equal or even a little higher on the phenom side.
too bad that 3dmark06 doesnīt give u an impression of real gaming performance.
In general, Phenom kept up like that.Quote:
Originally Posted by Hornet331
But I dunno, it seems like HWBox is getting... wronger and wronger?
p/s: CONFIRMED. HWBox performance is waaaay borked.
Their 3Ghz performance CPU score = 4136
If we consider ideal linear scaling to 3.8Ghz, we get only 5238. That's still WAY below Toppc's score (5682).
Probably HWBox has a 10% deficit on any compute intensive benchmarks?
pp/s: Could be Vista SP1, looking into performance impact as of now.
that 10% being applied to intel's quads too ?Quote:
Originally Posted by Macadamia
This.Quote:
Originally Posted by Tom128
If I can get ~260 HTT on a 920 I'll be happy.
actually i compared some of their benchmarks with some benchmarks from others i have seen. like coolice for example. the 3d mark one was the only one i found different if i had a benchmark to compare it to. their wprime one is legit although that benchmark is useless anyway. also their cinebench is legit too. maybe if they create another thread you can ask them to take a better look at the 3d mark scores since it seems their scores are lower than the other ones i have seen.Quote:
Originally Posted by Macadamia
seem like 12/09 bios is better:DQuote:
Originally Posted by Toppc
edited:my bad,cpu score are higher with nv card:p:
it IS vista ;)Quote:
Originally Posted by Macadamia
i hate both of you guys. :p::cool:Quote:
Originally Posted by coolice
Nop i am talking about PII and as dinos22 pointed score should have been 10711 , nothing more ,, look hereQuote:
Originally Posted by SkullCracka
http://i40.tinypic.com/2w6525v.jpg
Very impressive for non BE CPU's :up:
Can you test with default multiplier at 350MHz HTT = 5250MHz :D :D with subzero cooling definitely :cool:
i wonder how high the nb will clock probably. any info on this already? I've seen shots with 2,6ghz but dunno if those clocks are stable.. because with my 9750 that has mainly been holding me back using cnq during 24/7 use, whereas htt300 worked here on agena also:
http://www.abload.de/thumb/russmark33ghz1200mhz7oh0.jpg
i wonder if i could reach ht3.0/ht3.1 speeds (2,8ghz/3,0ghz) with nb/imc?
could be that the SM2/SM3/CPU scores are rounded, but the unrounded numbers are used to generate the final score.
EDIT: damn.. should use F5 some more...
It was stated somewhere in this thread that max. for this particular CPU sample is 26xxMHz NB clock.Quote:
Originally Posted by Oese
2.6GHz NB was stable. :)
yup vista eats quite a bit of performance on 3dmark. I wonder why the used win xp for that test. :p:Quote:
Originally Posted by dinos22
Also there is quite some fluctuation in points with a same clocked cpu and cpu score. Just look at the orb.
what is wrong with you man.Quote:
Originally Posted by Macadamia
http://img511.imageshack.us/img511/116/06uc9.jpg
if there is any performance loss at the deneb,the same loss will be for all the other cpus
As others said before, this has been always the case.Quote:
Originally Posted by Hornet331
Tony's comparison with a Yorkfield and a Agena B3 at the same clocks (think it was 3.4Ghz) shown the same score. They had same RAM speed/timings, same GPU etc, yet Agena B3 managed to get a higher GPU score while it had a lower CPU score, but it had the same final points as Yorkfield. But I think Phenom scales better with GPU's.
I did a small calculation a time ago, showing that Yorkfield scaled actually very bad with GPU's, while Phenom did it better. But well, this was often useless because of the lack of clocking on Agena B3's (although they got better after a while).
4.2G CPU-Z DONE
BUT CAN'T RUN ANYTHING
http://www.coolaler.com.tw/toppc/X4940/4200.JPG
Is this on air, water or something else?Quote:
Originally Posted by Toppc
To bad you can't run anything...
Would love to see some numbers at this speed.
Very insipring
AIR COOLINGQuote:
Originally Posted by Smalltimer
Xigmatek HDT S-1284C
http://www.xigmatek.com/images/hdts1284-bg2.jpg
Is that the old version with the heatpipes still far apart?
Or are they once again making a new version?
Can you reach the same clocks only with multiplier manipulation and if you can, please post some results with HT link oc and with multiplier.Quote:
Originally Posted by Toppc
As you see, it's not Achilles, so it should be with closer pipes position.
As you see, it's not Achilles, so it should be with closer pipes position. In Europe it's called S1284EE(Enhanced Edition), but as we see in Asia the name is different. @Toppc, could you tell the whole name of the cooler, what means this "C" at the end.Quote:
Originally Posted by Caveman787
Ram:27776MB , where can i buy so much ram? :cool:Quote:
Originally Posted by Toppc
new version!!Quote:
Originally Posted by Caveman787
HDT S-1284C
http://www.coolaler.com.tw/coolalercbb/HDT_S1284/2.JPG
http://www.coolaler.com.tw/coolalercbb/HDT_S1284/5.JPG
Appear on market for three months in taiwan
Should be a miscarriage of justiceQuote:
Originally Posted by Lastviking
Ah so s1284EE here is the same as s1284C there.
How were temps when you got it running stable overclocked?
I'm assuming you got a slightly lower overclock stable before you tried 4.2, just wondering because the s1284EE is a wonderful cooler and with how cool these things seem to run the temps should be really good.
Hi Toppc what is the highest stable oc you can get with a x14 multi? That is the same multi as the 920 uses.
all phenom II cpus have unlocked multis just it seems they are not using the BE after every cpu now since they are all BEs.Quote:
Originally Posted by bondhahnmrt85
still pretty impressive that you can get to 4.2ghz without raising the multi.Quote:
Originally Posted by Toppc
are you positive about that? (honest question not being smartass :p:) i have been under the impression the 920 had locked multi over 14Quote:
Originally Posted by roofsniper
i have read about it in the past but i can't be 100% sure.Quote:
Originally Posted by crazydiamond
thats dated November 18th. http://www.tomshardware.com/news/amd...ming,6612.htmlQuote:
The Phenom II X4 940 and 920 will also have unlocked multipliers.
im guessing that at least for now all amd cpus are gonna have unlocked multis.
Many thanks! That would point towards a 940 then... Mmmh gotta watch things develop until end of next week ;)Quote:
Originally Posted by Lightman
Yeah, thats what I'm wondering too. If they're all unlocked, I'll get the cheaper one to start, otherwise, its uptown for meQuote:
Originally Posted by roofsniper
There is a guy with the "Phenom II 920 goodness" thread, that has the 920 and I don't recall any multi OCing going on...so I don't se it being a BE. I might peak again tho.. I am board anyhow....(shhh...supose to be migrated a server);)Quote:
Originally Posted by nanohead
Sounds like this is just the difference between HyperTransport and the Intel FSB. We've always known that HT is a better I/O interface. Most people assumed the difference was only important in server apps but now it's obvious that there's a noticeable difference on the desktop too.Quote:
Originally Posted by Rammsteiner
TOPPC how you managed to climb up to 360HTT?? :up:
Voltage tuning or just simple matter of patience and pray?
Extremely impressive :shocked:Quote:
Originally Posted by Toppc
I know what my next mainboard will be :D
this looks to be a great board to pair w/ a 920
360 HTT SETQuote:
Originally Posted by Lightman
CPU VID 1.55V
CPU VCC 1.6V
CPU-NB VID AUTO
CPU-NB VCC AUTO
RAM VID 2.2V
HT VCC 1.2V
NB VCC 1.2V
NB SPEED 6x
HT SPEED 6x
MEMORY SET AUTO
PCI-E 100MHz