3200 Mhz stable on air default voltage maybe?Quote:
Originally Posted by HARDCORECLOCKER
*dreaming*
:p:
One problem though...my wallet isn´t that thick I can afford a pricey cpu.
Printable View
3200 Mhz stable on air default voltage maybe?Quote:
Originally Posted by HARDCORECLOCKER
*dreaming*
:p:
One problem though...my wallet isn´t that thick I can afford a pricey cpu.
holy :banana: :banana: :banana: :banana: that is :banana4: :banana: :toast: :stick: :confused: :p: :) crazy
:toast: cheers buddy, keep 'er clockin'
It's kinda scary if you think about it. This 3800+ is able to, at the least, compete with today's FX55 at less than half the price. If the San Diegos' silicon is truly the pick of the litter, their benchmarks compared to the Venice will be what SLI is compared to PCIE (Well actually not that drastic, but it's the only thing I could think of :) )Quote:
Originally Posted by HARDCORECLOCKER
:D Personally I expect the FX-57 to run 3,5GHz on single phase without any problems......, so we will see.Quote:
Originally Posted by s7e9h3n
:toast:
Okay now run some Prime95 goodness! :toast:
in cpu-z, you can hit f5 to save a bmp directly from the program instead of having to open photoshop/paint or whatever and pasting it. much easier imo :toast:
If only I could run my rig remotely from work :DQuote:
Originally Posted by WiCKeD
Didn't know that....I'll try that later on tonight. Thanks for the tip.Quote:
Originally Posted by ozzimark
:D O.K. - having an IQ only a bit above a banana myself so please tell again how it exactly work, open CPU-Z, hitting F5 and then ???Quote:
Originally Posted by ozzimark
:toast:
It saves the .bmp in the same directory ya have cpu-z.Quote:
Originally Posted by HARDCORECLOCKER
Named cpu.bmp if you have the cpu tab open.
What are you running server2003? It doesn't have remote management built in?Quote:
Originally Posted by s7e9h3n
If you freeze the machine your screwed, but I saw this device a while back that will reboot your computer if it detects a system crash. So, as long as you stick with Clockgen and a64 tweaker, you can run tests from work...
http://www.thinkgeek.com/computing/accessories/678e/
And even without this device You could at leats run some tests that aren't too crazy, until you freeze the system...
good lord s7e9h3n , that is really spectacular! excellent work. When ya get home from work, doa bit of benching, but then get some sleep man. We're excited to see what this Venice can do, but you don't need to be killing yourself over it.
Again, thanks a lot. :up:
:D Ah, I see now, THX mate.Quote:
Originally Posted by Caper
:toast:
You should thank ozzimark, didn´t know about this before he mentioned it.Quote:
Originally Posted by HARDCORECLOCKER
Pure luck that I noticed the screenshot was saved in the cpu-z directory, or is it so that I have an IQ above a banana...maybe like two?
:D
worked for me, thanks for the tip, btw anyway to save to jpg directly?Quote:
Originally Posted by ozzimark
here's my screenie, if anyone wants sandra or w/e benchmarks just ask :p:
(can't use my a64 system since I'm waiting for replacement 3500 and VX)
http://www.xtremesystems.org/forums/...id=26769&stc=1
So let me get this straight, the following was done on air ???!!!!!
http://img164.exs.cx/img164/2226/patched9pb.jpg
:slobber:
Perkam
No. That one is playing in the cold.
nope. not yet at least. i would actually prefer .png because it's smaller and lossless :toast:Quote:
Originally Posted by S0nic
and conrad, how cold?
Check back a few pages. I think he's playing in this temperature range.
No, I believe this was the most on air:Quote:
Originally Posted by perkam
http://img57.exs.cx/img57/5921/10x2955oq.jpg
thanks, i somehow missed that while i was at class from 12-4 today. people post so fast :slobber:Quote:
Originally Posted by conrad.maranan
Now don't get me wrong - I'm into testing/benching my systems just as much as the next guy - but whoa, that is truly going a bit overboard. And besides, if you're gonna test the capabilities of a system, you're not pushing hard enough if you're not borderline BSOD :DQuote:
Originally Posted by J-Mag
Thanks conrad - I was busy contemplating remote-benching for a minute there - :DQuote:
Originally Posted by conrad.maranan
Also, here's something odd to throw out to you guys. This cpu, for some reason, doesn't like it's default multiplier (12x) at these higher frequencies. It requires much more voltage at 12x to run any kind of bench. On the other hand, 11x seems to be much more of a "sweet spot" @ 3.0+Ghz for the chip. It seems to run more smooth and with less heat and vcore than both 12x and 10x.
is that TCCD you have? can you pull some max mem runs for use so we can see this new mem controller in action?
thanks
if you OC remotly you will be my master xD omg remote OC and bench ppl are getting so far
bh5 2x256 i think
Wow... 37% overclock on Vapo!
Is it me or are those results slightly underwhelming?
2x256 G.Skill 4400 L.E.Quote:
Originally Posted by stealth17
I don't have a screenshot of it right now, but @ 1:1 11x300FSB Cl2.5-3-3-7, sandra reports buffered bandwidth around 8600mb/s Int and 8600mb/s Float. I'll see about maxing it out later tonight....
Underwhelming in what sense?Quote:
Originally Posted by bugeyes
remember thats only loke -12c load...someone with -50 or -100 load will see more improve im sureQuote:
Originally Posted by bugeyes
EDIT: I'll see about maxing it out later tonight....
thanks!
that higher than you could hit with a winnie still?
download grabclipsave... set it to take Jpegs at 70% at 800x600.Quote:
Originally Posted by s7e9h3n
this takes extremely fast screen shots that are extremly easy on the system.
press print screen and it will instantly take a shot and save it where its set to save things by default.
http://boumchalak.net/index.php?opti...ry&filecatid=1
using Bitmaps at insane speeds just makes the system crash.
the trick is to find a way to take pics... before the system crashes ;)
mmm digital camera?
Remember, he's using a 3800+ so his percentage increase over default will be a lot lower than someone pushing a 3000+ to the same frequency. His results are rather worthy of oohs and aahs considering the chip is running neck and neck with the FX-55 he posted screenshots of earlier.Quote:
Originally Posted by bugeyes
Which store is that? I can't find it... :stick:Quote:
Originally Posted by Caper
In that I was expecting more megahertz with strained SOI and the smaller die... when cooled by phase change.Quote:
Originally Posted by s7e9h3n
I guess it's just too early to make any conclusions with just one sample of a high end core, lets just hope the slower speed grades will OC to similar MHz. If a 3200+ can do 3gig with stability I will be happy!
send that cpu to fugger :D maybe with his cascade can OC better
sure that cpu has strained SOI ???? i asked that and nobody answer me that :stick:
Thanks Kunaak, this just may come in handy ;)Quote:
Originally Posted by Kunaak
it is actually unkown at this point whether or not Venice makes use of SSOI. It is assumed that San diego will I believe.
we will find out if the San Diego starts hitting 4ghz with ease :rolleyes:Quote:
Originally Posted by trakslacker
although, maybe thats why the chip doenst much benifit from phase-change :confused:
It is also possible to up the FSB in clockgen, press Printscreen, lower the FSB and then save a screenie.Quote:
Originally Posted by Kunaak
Well, from the looks of it, Fugger seems to be busy currently with his Intel platform. And "maybe with his cascade can OC better" would be quite the understatement. (Although it wouldn't be an enormous amount over what's capable @ ~ -20C - ~30C.)Quote:
Originally Posted by leviathan18
I'm not sure if the Venices are based on sSOI. They may be just plain SOI. Check my post with the shot of Cbid. I know some things in it are faulty, but it does have an indicator for process type. I think it said SOI.....
no info of that in amd websites? i tought that sSOI was reserved for FX lineup but not sure on that.
i think they said they will need sSOI to scale further with the athlon 64 and cooler too.
I could do that, but I'd feel like I'm cheating :DQuote:
Originally Posted by Ref
LOL i was just joking with the fugger thing is your cpu and i think you should take the **** out of it you alone...
mmm digital camera is not possible? i think is safer
I wouldn't trust CBID right now, if it didn't even recognize the core correctly, its outdated.
Also there's no way to tell whether the CPU has sSOI via software.
exactly is just a manufacturing process i dont think soft can say if has sSOI unless they associate sSOI with certain core
Only 20 more posts s7e9h3n and you can sell your CPU here!
:D:D:D
Awesome CPU
s7e9h3n if you could answer a quick Q for me: What's the stock vcore on these chips? 1.35? 1.4?
I've had boards in the past where the CPU-Z vcore reading and BIOS vcore reading didn't match up, and the CPU-Z reading is usuall much lower than the chosen setting in the BIOS, so I just wondered what vcore this chip is "supposed" to be running at by default. Thanks!
Dang nice overclock
1.4 and are you sure you're just not experiencing vdroop?Quote:
Originally Posted by sabrewulf165
Looking good so far, s7e9h3n :up:
Cause it's working well under phase I am quite happy. I hope San Diego will do as well ;)
BTW. Does your nick mean: Stephen?!? :D
It is not considered as a cheat. A lot of people do it ;)Quote:
Originally Posted by s7e9h3n
And if you like to save the screenie in Kunaak's way(800x600 70%), I see no reason why you shouldn't take this screenie in a full quality.
Yes, all 90nm parts have at least 1st gen strained silicon. These may (or may not, we don't know for sure yet) have second generation strained silicon, known as "DSL" (Dual Stress Liner).Quote:
Originally Posted by leviathan18
Interesting that your CPU-Zs show 1.36v on a lot of the "stock" runs with air cooling. Are you deliberately undervolting slightly, or is that just happening despite being set to 1.4v in the BIOS?Quote:
Originally Posted by s7e9h3n
Actually it's set to "default Cpu VID" (or something like that) in bios. 1.4 is default vcore for the venice. 1.36 is probably due to the motherboard undervolting a bit. Anybody wanna send me an Ultra-D to test with this cpu? :DQuote:
Originally Posted by terrace215
It would cost less for postage if you sent me the cpu to test it in the dfi ;)Quote:
Originally Posted by s7e9h3n
Yep, I have a complete retail box Ultra-D on the shelf.... you wanna send me the 3500 to test I will send you the DFI to test. LMK!!Quote:
Originally Posted by s7e9h3n
:banana4:
Looks great! :toast:
It would be nice if AMD had a processor spec list like Intel's Sspec page on their site.
THanks conrad. OMG i posted only 2 hours ago and its 2 pages ahead :eek: Will check them out, thx for 2945 linky and OMG Conrad when'd u go mod ??? And lets some benchies in ur sig man !!!Quote:
Check back a few pages. I think he's playing in this temperature range.
I still wanna see some 3200+/3000+ benchies though before I'm completely conveniced :D .
Perkam
I'm loving :buddies: the looks of this Venice core. Hopefully the rest are as awesome as this chip.
P.S. Are you going to test the 3500+ ?
Very nicely done so far s7e9h3n :toast:
Would really like to see how the on-die mem controller does on that puppy ;)
Maybe a "tightest timings TCCD, 4 DIMMs" with you FX and then with the Venice? (you could do it with 2 DIMMs, but the difference would probably show up more with 4) Pleeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeease.
Thought part of the whole point of doing Venice/SanDiego was to implement the IBM version of Strained Silicon.... aka Super Strained Silicon. Along with SSE3 and better memory controller.
Well, the cold bug appears gone, SSE3 was verified, so that leaves the mem controller :) As far as the SS question... something fixed the cold bug... and from the temps at volts, it appears they've cut down further on the leakage current :D
dose the coldbug have any relationship to the memory controller?Quote:
Originally Posted by EMC2
Yes. This thread is definitely popular. Since s7e9h3n first posted yesterday just after 1:00PM, this thread has been hit almost 38,000 times. Notice that the moment s7e9h3n puts up a screenshot, the posts start accumulating. Everyone has been waiting a long time for this. ;)Quote:
Originally Posted by perkam
As for moderating: I woke up yesterday and noticed Kazoo and friends at my door. FUGGER was in the background next to his pimped out Benz S600 sitting low on 24" Lowenharts wrapped in Pirelli P-Zero rubbers. chilly1 instructs him to pop the trunk and reaches for a Louis Vuitton duffle bag. They both commence to walk toward my door and post up next to Kazoo. chilly1 drops the bag and a fat stash of bundled cash makes its way out of the designer pouch. Kazoo hands me a badge that reads XtremeSystems Bodybag Crew and mumbles, "It's time to clean house." FUGGER looks me in the eye and says, "Make sure you do a good job." I put the badge on, shut the door, and log on to XS. The end. :cool:
Unknown, but unlikely, could be an initialization problem most anywhere in the chip... I simply meant that the main Q left to be answered was what is the improvement in the mem controller on the Venice :)Quote:
Originally Posted by bugeyes
--- Conrad ---
:rotf:
SERIOUSLY OT:
^Dude, that sounds so cool. No, seriously. :cool:Quote:
Originally Posted by conrad.maranan
Uhh... lower speed rating Venice action? I wanna see a 3000+ hit 2.9GHz stable on air. Please? Anyone? AMD, are you there? April is so far away...
hmmm...2.9ghz would be 9x322 so someone with Venice + Gskill or Plat2 could do it, though it would be easier with a 3200's 10x multi : only 290 fsb required which many tccd can do.Quote:
Uhh... lower speed rating Venice action? I wanna see a 3000+ hit 2.9GHz stable on air. Please? Anyone? AMD, are you there? April is so far away...
Perkam
322x9 w/ a 166Mhz memory lock would suit me and my UTT sticks just fine.
I think you forgot about dividers here ;)Quote:
Originally Posted by perkam
322x9 divider166 = 263MHz --> good for some 6-2-2-2.0-1T BH-5 action
322x9 divider180 (on some boards, like DFI NF4) = 290MHz ---> great for some good clocking TCCD @ 7-3-3-2.5-1T
so, I think 2900MHz on 3000+ is still doable ;)
:D But still 322 2.5-4-3-7 is doable on G.Skill memory (at least mine does 325) with OPB's (holy) timings ;)Quote:
Originally Posted by bachus_anonym
http://www.sbs.o.se/default.asp?artnr=ADA3500BPBOXQuote:
Originally Posted by Pjotr
I expected more. However, first winchesters SUCKED and later ones got better ;) So I think we can expect the same thing to happen here as well. Venice doesn't suck to begin with, so it will hopefully get close to 3.2 on AIR by Q3-Q4.
For non-believers remember FX-51 needing serious phase to get to 2.8 and some FX-53's getting there on air ;)
I would love to see the 3500+ results now.
Quote:
Originally Posted by perkam
322x9 will slaughter 290x9, its no contest! Plus the 3000+ Venice is cheaper and if im forced to get a new mobo, I might go msi k8n then run like 350x8=2.8GHz which is doable on a venice and run my ram 1:1 thanks to its ondie controller :slobber:
Hey gf4, remember when i told you that venice would be 3800 and 3500, then filter down? And you said no, you were sure that they would come out first at 3000, 3200? Remeber that?
most people say 3000+ thru 3800+ venice is comming. its on amds roadmaps too......
Nice job Steven :toast: :toast: :banana4:
Q If this venice core is able to clock easily to 2.8 plus on air , does this mean that the fx 57 could actually be clocked at 3 gig ? since it is amds best silicon(that we know about) why not give intel a serious scare? come on AMD instead of suing intel (a possibility according to www.theinquirer.net) hit them where it hurts most !!!
Show them whos the BOSS!! :toast:
First of all respect to you steven for clocking a Venice that high, but....
what is up with using SuperPi anyways? It's like whoooh I got 25seconds 1M, So....and...? :confused: :)
You compared a FX55 (291x11 IIRC) to the venice with superpi, and some Sandra benches......why not some benches in real life Games...catch my drift? :)
I mean 95% of all people that have high end systems use them for gaming...so what about a comparison between the FX55 and this Venice?
Facry
HL2 and CS:S
UT2004
Doom3 etc etc....
this way we know what the real life gains are...not just some synthetic (sp) benchmark.....it's not alot of work (if you still have the FX55 setup up and running)
HL2 bench : http://downloads.guru3d.com/download.php?det=927
Farcry Bench : http://downloads.guru3d.com/download.php?det=830
Umark (UT2004) : http://unrealmark.net/
no offense btw --> :)
I agree with Tim, some 3DMarks would also be interesting - especially 2001. Also prime95 ofc, but then we need to give him some time. Let him hit the definite ceiling first and then go from there. Keep it up!
Yeah, let's post some FPS result for game will be more interesting.
I'd like to see some Prime95 screens now, superpi 1m doesn't say much about stability.
The difference between superpi- and prime-stable can be HUGE
I'd like to know if this cpu can Prime at 3ghz.
To bring you latecomers to the thread up to date--Quote:
Originally Posted by krille
He does not have a good video card to test with, but only a crappy PCI card to see how the CPU clocks.
I believe he has an AGP card and his mobo is PCI Express?
He has a 9800 pro and an agp board, but that doesn't mather that much.
The 3dmark score will be just below or equal to tha FX55 at the same speed.
Stability testing is more important !!!
Quote:
Originally Posted by TysonM
Yeah, sorry, I forgot. (Had read the entire thread though.) It'd still be interesting with 2001 and Prime95. Prime95 being most important. Anyway, we have to separate wishes from reality, give him some air.
Ladies and gents:
Everyone needs to chill here. s7e9h3n has been gracious enough to show his results with the new chip, and ever since he has been bombarded with requests, complaints and nags. It is really getting on my nerves - this thread is clogging up so fast it's difficult to track s7e9h3n's postings, which are the only ones that matter since he is the only one that has the chips, please remember this!
For the confused: the "Venice" cores are the Revision E (or E4 stepping) 512KB L2 Socket939 Athlon 64 chips. They are produced on a 90nm strained silicon-on-Insulator process and are SSE3 enabled. IPC simply describes a chip's performance on a per-clock basis - if chip A is faster than chip B at 2GHz, chip A has higher IPC, regardless of how or why. A die shrink does not inherently increase the per-clock output of any semiconducting device, die shrinks increase "performance" in the way of better yields (for a given frequency) and higher frequencies (for a given yield), manufacturers sometimes refer to performance in this way.
If you have further questions or commentary about the chip's architecture, IPC etc. those should be directed to another AMD thread about the Rev. E chips, of which there are many. If you wish to argue with the above paragraph, PM me, do not post your replies here.
Well, because this isn't arguing, just a correction, i'll post it.
Venice Core's have the E3 Stepping, Not E4. E4 is reserved for the San Diego Core'sQuote:
Originally Posted by matt9669
And do we know for certain that Venice has strained silicon? We can only speculate at the moment
Why perma ban for a stability or benchmark? I, and many others, would like to see the stable potential of the chip, not a spi 1m at 2.9. A lot of winnies can suicide that, so it is important tahat we can see a 8hr p95 at whatever level possible
matt9669 that was way out of line. There is no legitimate reason for PERM banning someone that asks for a benchmark. If steven is unhappy with the comments in this thread or doesn't want to be asked questions then let HIM say it. He doesn't need you to say it for him. If you want to find only his posts then search this thread for only his username and read them. Stick to the news next time and leave the moderating to those with better judgement...
Well said.Quote:
Originally Posted by Techmasta
Yeah, I agree people should stop begging and nitpicking (people should assume stephen hass a life outside of OC'ing) but I will put my neck out there to say that you are out of line Matt. What we can't discuss anything you posted without PM'ing? I guess it's all law based on your mod status?
Other mods disagree with you on the IPC thing. Some will go and say if a winchester outperforms a newcastle both at 11*200 the winnie has higher IPC, basically using the whole MHz * IPC = Performance relationship. But IPC can also be discussed more in depth and slightly differently. You can say something like an Athlon 64 can do 6 arithmetic operations per cycle and 3 floating point... for example. This does not change newcastle-> winchester-> venice. When you look at it this way, SSE3 and such, while boosting performance, is not a boost to IPC. Every little thing that boosts performance without MHz is not always considered increasing IPC for those looking beyond MHz * IPC= Perf. The errata consequence model discussed was good info on reasons for perhaps higher performance. I haven't looked into the validity as a whole. But please consider all facts before using your divine powers to shut down discussion.
I think GF sometimes says higher IPC when it would be more correct to say cpu x is faster than cpu y clock per clock. The way it's misused, it wouldn't be far fetched to say an FX has a higher IPC than non FX. While cache equates for more performance, not all of us would say it has a higher IPC. An FX has the SAME IPC. It just has the other advantages that come with more cache, like reaching out and waiting for system mem less.
Quoted for agreement.Quote:
Originally Posted by Techmasta
Now this thread is getting really off topic. Could a mod just remove all posts other than steven's and those on topic?
Thanks
You guys are right about the 1m super pi test doesn't really show stability.
The 16m test would be better, or of course Prime 95.
Anything like that would show the stability, as well as if he could mark down the temps at full load on these tests.
Stephen: how about testing the 3500 now, as it's more of an affordable chip for us humble people where as the 3800 is a few hundred notes more :)
Andy
bah, it's an open forum. the manner and arena in which this is posted clearly invites comments, questions, speculation, and yes even in some cases stupidity. clearly not a bannable offense simply because it gets on your oh so sensitive nerves. :stick:Quote:
Originally Posted by matt9669
great results s7e9h3n :toast: thanks for sharing....
Why not? SPi is enough. Who cares about stability here, there are no rewards for stable systems. That's why this site isn't called "xtreme stable systems.org."Quote:
Originally Posted by zA.Gosu
ppl DO have lives outside of providing requested benches on 0-day hardware, give the guy a break eh.
be grateful for what u have seen so far ....when he gets a firmer grip i'm sure he will do some stability posts :)
:D Subscribed, and the definition of stable is also depending on each person's personally view, some poeple definate their rig as stable when only gettin' a screenie and others say it is only stable if running 24/7 prime95 for 3 month........Quote:
Originally Posted by HaLDoL
:toast:
Sure there is, actually being able to use the computer i just paid a 1000$ for is my reward.Quote:
Originally Posted by HaLDoL