Agreed. The only way i would give up Intel for BD is if it beat a 2500K in IPCQuote:
Originally Posted by Megacharge
Printable View
Agreed. The only way i would give up Intel for BD is if it beat a 2500K in IPCQuote:
Originally Posted by Megacharge
It won't, nobody paying attention ever thought it would. So you can stop waiting and go back to playing with your Intel rig.Quote:
Originally Posted by Pestilence
First off save your attitude for someone else. Im waiting for BD so I can build my fiance a 4100 machine.Quote:
Originally Posted by Solus Corvus
That doesn't mean what you said isn't illogical. Performance is a combination of clocks and IPC. But you are saying that performance doesn't matter and you personally will stick with Intel if BD IPC doesn't beat SB. I'm saying it is illogical but if that is your criterion then you already have enough information to decide which product to get for your own personal rig. It was never in doubt, Intel is aiming for max IPC and AMD is aiming for "knee-of-the-curve" IPC.Quote:
Originally Posted by Pestilence
I go by what we know thus far. Not magic pixie :) . All leaks so far point to the same unfortunately. Yes they are all from Far East,but so what? If we get better data points maybe we can discuss it,so far this is all we have and it doesn't look very promising. And xsecret is not the only guy who is not an optimist (and he has real hardware like some other guys). Don't get me wrong, I wish Zambezi to succeed more than anyone else.Quote:
Originally Posted by Oese
Yes ,there is a chance all this is a "game" played by paranoid AMD: seed everyone(press too!) with sub par samples,play the frequency game card,price it very aggressively etc. But until this is confirmed with hard facts I choose to be realist. And what we know by now not only shows sub par (to Thuban) integer performance,but sub par SIMD performance too.. Yes it makes no sense to me either but that's the way it is.
someone at another forum pointed out that all the leaked cpuz 1.58 screenshots of BD lookQuote:
Originally Posted by xsecret
different than this WR cpuz screenshot,is this maybe because the leaked cpuz SS's are fake
because i looked at the FX8120 from VRzone and the chiphell ones and they are different.
EDIT:never mind i think i figured it out,with only one
module in use the cache changes in cpuz
with 8 core/4 module its
8x16
4x64
4x2048
8
with 2 core/one module its
2x16
64
2048
8
Firstly, congratulations to Chew, Macci, Et al for the excellent overclock. It sounds like these bulldozers can really clock easily. Now I'm really curious to see what kind of performance they have at "daily driver" stable clocks - interesting times ahead.
Some thoughts about this thread:
1) "It can't be worse because X"
Several people have said essentially that it can't be equal to or slower than previous products because that would be bad. That's a logical fallacy called an appeal to consequences. It's summed up as "If x is true then that would be bad, therefore x isn't true". Unfortunately there isn't a logical basis for this, it's all emotion.
To the obvious retort, "well then they could have just shrunk PII instead", I say it isn't that simple. What should AMD do if GF's gate-first process really does have issues? Let's say AMD got most of the bugs worked out in the early ES chips only to find out that GF probably won't be able to hit original clock targets and stay within TDP? It's far too late to do something like re-engineer thuban for 32nm. What they would do is release what they can achieve and tread water until the process is improved or a new one can be implemented. 8 cores looks better on the box than 6 cores. Not to mention the server world would be lukewarm to a warmed over thuban. Fanboys would be disappointed with either outcome anyway, but they don't care about TDP so BD would be more interesting than 32nm Thuban. Releasing BD is the best business move. It makes AMD more money right now because they know (or should anyway) that marketing can be just as important as having a top chip (Remember Intel in the P4 days).
2) "If it doesn't beat SB AMD is going out of business."
Nonsense, this is an emotional reaction. People have been predicting the demise of AMD for years. Remember Shintai predicting that, was it ~3 years ago, that they would be out of business next year? They have strong products in several market segments and their fundamentals don't show anything to be particularly panicked about. Interlagos should help with their margins a bit and give them a little breathing room. Enthusiast sector chips are still a question mark but even if they don't knock it out of the park on the first swing they will get more chances at bat with further product evolution.
3) "If AMD doesn't deliver this time I'm quitting them for good."
That's another emotional overreaction. It makes about as much sense as breaking up with your girlfriend if this one last batch of silicon chip cookies doesn't turn out perfect. And by girlfriend I mean the girl you are stalking. It makes sense if a product isn't up to par you will buy the competition's offering. But that doesn't mean you shouldn't consider any of their future products. If they do eventually make something that competes, or even dominates, it wouldn't make sense not to consider it. Buy what offers you the most performance in the applications you run for what you are willing to pay. Brand loyalty etc is illogical.
4) "IPC is low so it sucks."
Performance is a combination of IPC and clocks. IPC doesn't mean a whole lot by itself. What we should care about, and what I am personally waiting to see, is what kind of performance it has with a 24/7 stable overclock. That should be our basis for comparison as enthusiasts: BD average OC performance VS SB average OC performance in the apps we use.
5) "Chew's car"
Personally when I modify a car I try to play off it's strong points, unless it has some really glaring and easily fixed faults. When going for lowend grunt I'd choose, and did, a muscle car with a big block. Then further mods to play off that strength in lowend torque and accentuate it. For twisties I'd go with a 4 cylinder and play off it's ability to rev. If your car has good midrange and topend performance I'd get a higher numerical differential gear + engine mods to help it rev. The differential will help with lowend acceleration and get you into the high rev powerband quicker. This whole IPC vs clocks talk reminds me of the displacement vs revs talk in the car world - but performance depends on the combination of both and what you are trying to achieve. :)
Also I'd like to add...
If L1 WT was the reason for high frequency scaling, then they should have simply applied it to K12 (Llano) and clocked it to hell, no?
6 Stars cores on 32nm high performance process doing 4.6-4.8 Ghz maybe?
I wish I could say I expect more out of Bulldozer...but 8.4 Ghz is too exciting right now XD
No offense if anyone here made this claim, but I cringe whenever someone states something so ridiculous.Quote:
"If it doesn't beat SB AMD is going out of business."
Indeed. First the assumption was that it wouldn't clock high, now that it is proven that it does, some folks (unfortunately) are finding reason to still be disappointed. Oh, well.Quote:
I wish I could say I expect more out of Bulldozer...but 8.4 Ghz is too exciting right now XD
The worst part is people are doing it to themselves. I'm not a Buddhist but I think Buddha had a good point about suffering coming from desire and ignorance. By desiring something you can't have you are provoking angst within yourself. And ignorance is not seeing reality as it truly is. By making false assumptions and ignoring facts we feel anxiety later when reality doesn't want to cooperate.Quote:
Originally Posted by PerryR
Best to be patient and let it come to you and accept it for what it really is, whatever that may be. Good or bad.
True. I'm personally not going to be disappointed either way.Quote:
Originally Posted by Solus Corvus
BTW, I'm really friggen excited now:
http://www.brightsideofnews.com/news...er-system.aspx
Nice, and it says available "shortly".Quote:
Originally Posted by PerryR
So maybe the release date for BD will be September 19th as AMD said it would after all.
The WR Overclock screen had an image of some software I'd not heard before on it... naturally I looked for it, almost to no avail ;)
http://wwwd.amd.com/AMD/SReleaseF.ns...ToolsbyOSAMD64
Good gravy there are some nice utilities on that page! PSCheck is like K10Stat, but by AMD (and visually supporting X6 chips, among others like Trinity, Krishna, Orochi, Llano...)
System Access makes me wish I knew more about hardware access codes and the like, because it seems like it'd be the ultimate tweaking program I've always wished someone would release!
I'd go on, but I know you all are capable of reading lol
You tease! I noticed yer signature... can we expect spilling of some beans?Quote:
Originally Posted by 2good4you
Time Magazine - AMD Breaks 8GHz World Record with Daring CPU Overclock
Read more: http://techland.time.com/2011/09/13/...#ixzz1Xtk2QGQX
AMD going $7+
http://www.marketwatch.com/story/amd...cpu-2011-09-13
And you, you're paid by AMD for sayin' such bullits or what ? :confused:Quote:
Originally Posted by 2good4you
Are you aware that AMD has sent confidential datas to some press ? Do you really think that data directly from AMD are indeed faked data ? :confused:
Good find! Have we known about these tools and/or these pages? Would be nice to have access to all of them.Quote:
Originally Posted by Formula350
Does anyone else get a "Dr. Who" vibe from this guy?Quote:
Originally Posted by Olivon
Olivon: I know you've been around for awhile, but your location almost exactly matches his, as well as your deleted sig "Doc???@???".
Your comments sound very similar too!
I apologize if I'm wrong, but why exactly do you keep posting in this thread if your so sure BD is a fail?
Olivon
And you, you're paid by INTEL for sayin' such bullits or what ?Quote:
And you, you're paid by AMD for sayin' such bullits or what ?
Are you aware that AMD has sent confidential datas to some press ? Do you really think that data directly from AMD are indeed faked data ?
Are you aware that confidental data AMD has sent to some press are under NDA? Do you really think that data directly from OBR or Chinese guy can't be faked data ?
I hope you like it:D
I see that some of the "leaked" benchmarks have model number FX-8130P. Is the FX-8130P model number going to have retail availability or is it ES only?
Direct data from AMD? :confused:Quote:
Do you really think that data directly from AMD are indeed faked data ?
More in this link: http://www.nordichardware.se/test-la...srekordet.htmlQuote:
NordicHardware: "It says you reached frequencies "well above 5 GHz" with only air and sub-$100 water cooling solutions. Mind going more into detail about your ventures on air and water?"
Sami Mäkinen: "In terms of CPUZ MHz I’ve seen virtually every CPU reach over 5GHz when using a good air/watercooler. I believe the highest result I saw with aircooling was around 5.5GHz CPUZ. This CPU hit around 7.8GHz on LN2."
Yup, maybe it's just me but I've been finding his and xsecret's posts a bit suspicious. Something just seems a bit weird with those guys.Quote:
Originally Posted by Daveburt714
I hope the 5 Ghz Air/water is true...that would really be awesome...