Will HKMG make AMD superior? - Page 2

You havnt read everything Ive posted, nor what Ive read. Those 2nd parties did have access, they did play around with them. Theres mobo employees whove done quite alot with them as well. Ive been doing alot more reading than you, and the sources are VERY respectable. Its only a matter of time before this all happens anyways. I dont think Deneb will match up in IPC, but may turn out to be a better oceer, so including pricing, its another alternative

Ah, numbers keep changing. Yesterday everyone said "past 6GHz" on LN2. Now you say 6.3 without a link. That was a single commercial demonstration and for some reason they gave a partial NDA.

On air, Nehalem more than keeps up. Most air OC's on XS are hitting 4.2-4.5GHz; there's even a 4.4 with stock cooling, but without HT. The problem with Nehalem is the low ceilings at cold temperatures. Some chips have a cold bug around -60C - bad for LN2 but good enough for dry ice. Others work fine with LN2/cascade. Some clock to 4.7 tops, others closer to 5.2. And you know XS folks are resourceful. None of the chips are showing great scaling with cold.

To win the OC race but only under LN2/cascade is a symbolic victory. There are only a handful of people who use LN2/cascade for extended periods of time. For everyone else, Nehalem stays on top of Deneb in that even if the frequency averages a bit lower, the IPC more than makes up for it.

Thats true, and so far thats what we do know, and you know it was 6.3, as I think it was agina Andreas who confirmed it. But yes, a hollow victory true, if on air it isnt better, and still including pricing, you cant escape that, have to look at the whole package, price/perf. If price isnt a problem then alot of the time more exotic cooling solutions are used anyways, which could also favor deneb, so to most price perf, as you know. But thats so true, if its only the LN2 numbers where it rules, then it is a hollow victory.

Its still early, and dont misread my comments, as alot of them are done just to ascertain the truth about these chips, as alot of people just dont believe it yet.

I never doubted the clocks on i7, it was and still is its IPC, especially comcerning gaming. Allll Im saying is, these clocks are real, as too many sources have agreed to a tee as to their performance/clocks. The IPC is what weve seen in Shanghai, ho hum, nothing earth shattering, itll all come down to clocks/overclocks and pricing, and that you can take to the bank, if they havnt closed for good by now heheh

PS If youve read about Intels planned % of i7 vs their current lineup in production, well be seeing the FSB models for awhile yet, so, as for overall success of Deneb and AMD, itll be competing mainly against those chips, and should show decent competition, especially IF priced right

Reading the link now. The 5Ghz, if i recall, was still playing th Crysis demo. The NDA was for top air only speeds as well, or so it seems, as it seems everythings been rounded, and the 6+ was a no no From your link :Beyond that, from a clock-for-clock performance perspective, optimizations have been made to the CPU pipeline, including improvements to branch prediction, pre-fetch mechanisms and TLB (translation look-aside buffer) optimizations. In total, along with larger cache sizes (now 512KB L2 cache per core with a 6MB L3 cache) and higher clock speeds, AMD claims the new Phenom II X4 should offer a 30 – 40% performance increase over all.







TY for the link, nothing I havnt read except the AMD claims . Im thinking those claims include power savings in the averaging as well, as a 15% clock jump + say 7-9% in IPC doesnt get there, so must be power as well

I'm having a severe flashback regarding the "40 percent improvment" note from AMD.

Now now, context please, thats their claims, not mine, mine are 15% clocks, 7-9% IPC. In order to get the 30-40% claim AMD has made, youd have to add in the power savings as well, whic h weve all seen with Shanghai

Its not unreasonable to see 7-9% IPC with a shrink. We all know theres a 3Ghz Deneb on the way, adding what we know about Shanghais power savings, it all adds up. Simple

If you consider power savings performance, thats their shtick, and a failing one , I think. It should be kept seperate, and Ive never included them, as theyre much more important to server than DT anyways

If Im wrong, if theyre wrong, and AMD has somehow deceived us all again, well, guess we will all feel the loss. The same holds true for me, as Im not a AMD fanboy, but Id still feel the loss of such distrust. AMD cant afford this, I think they know this, which is another reason I believe it this time around, because if theyre wrong again? The impact will send them down the drain, and like Ive said, we all lose except for the short sighted Intel fanboy in the truest sense of the word. This isnt life or death, its fun, and if thats gone by deception, well shame on me I guess

if you think that is cool there is a report that AMD Phenom II can hit 6GHz on LN2 but we'll wait to see it. http://www.nordichardware.com/news,8389.html

Im not flaming anyone inthat statement, it only applies if it sticks, and thatll show up on those that cant let it slide off

Even if that's true:

1. For 99.999999% of PC users it's meaningless because they'll run at stock speeds or mildly overclocked.

2. You're talking 8 threads at 5.7GHz vs 4 threads at 6.3GHz. I thought clock speed didn't matter anymore and performance was what counted? Or was that only true when AMD was providing more performance per clock than Intel?

Personally I couldn't care less about LN2 overclocking, what does concern me about Nehalem is some of the power-consumption figures at stock speeds, which are well beyond what I'd want in a desktop system. Wake me up when Joe Sixpack can actually buy a 6.3GHz Deneb, and it's faster than a 5.7GHz Nehalem, and run it either at a sensible power level.

Thats my question too. I remember when Athlon 64/X2 came out. It was about it being better performance per watt and per clock. Now that never comes into play.

Its like a one sided war. But meh what can you do.

My opinion is still that this is just hype. Until I can see everything about Phenom II that was stated as true I wont hold it true. Last time it was held true ppl got so let down. No reason to let that happen again, right?

From my understanding, using both SOI and the new immersion technique, its strenghthened the channels, like youve said, but cant carry the voltage Intel does because of the gates, which is where HKMG comes in to play. Once applied, the gates will also be beefed up and maybe see better numbers , or much better numbers than Intel, which, in the end would allow for higher clocks and lower power draw. Thats the way I see it

Actually there are a few factors here. 1. we do not know what other material IBM is planning on using. From what we know Hafnium is the best one to use as per Intel who had been researching this for much longer than IBM or anyone else. It turns out that Hafnium allows to go smaller without it allowing more leakage but also does not change the resistivity, which is a big deal in electronics.

Also if AMD does do HK/MG then they would more than likely have to have a new lithography and immersion tech because HK/MG is different in a lot of ways to SOI.

What this really means though is that once AMD does start to do this they will not have it perfect up front. It will take a bit of time to do it good unless they wait for 32nm and IBM to have a bit of experience with it. Remember Intel has been working on it since they started 65nm so they have had quite a while (about 2+ years) to actually work on the process of implimenting HK/MG. We could see even better results from it in future CPUs if Intel keeps working on improving it, which they will.

If AMD rushes it the improvements might not help much. But if they take their time they may see what Intel saw at first, 20% better switching performance and 20% less leakage.

But only time will tell. I doubt they will rush it to 45nm. I think they will wait till 32nm. That would be a better idea.

I still find it kinda ironic how AMD said SOI was the way to go yet they are going to end up ditching it for HK/MG.

Not really knowledgable about SOI and HKMG, but from reading the above posts what effect will overclocking have on processor life? It's great that Deneb looks like a good OC candidate, but is upping the vcore going to have a more signifigant negative effect on 45nm SOI chips?

Actually thats not true. AMDs voltage is about the same as Intel. My Q6600 only uses 1.25v for 3GHz. Most Phenoms are at about 1.35v for 3Ghz. And thats with both at 65nm.

No matter what process you use too when it gets smaller, the room for voltage gets smller. Thats why I doubt that 1.55v will be good for a Phenom @ 45nm because its channels are not as wide as 65nm was thus voltage that high would be horrible.

Right now Intels 45nm only uses about 1.1v or less for their chips at 3GHz. Thats a nice drop. And their top is at 1.35v compared to 1.5v for quad 65nms. Hell with the E8600 can hit 4GHz on air with less than 1.2v in some cases.

OCing always lowers the CPUs life but if its moderate OCing (say to 3GHz) then it wont be noticeable. Even if you OC it a bit mroe it will still last about 10 years. Maybe longer. Only when you do insane OCing does it tend to lessen the life and the worse the leakage (depends on the process) the faster it will die with higher voltage.

Going from 1.1 to 1.35 is about 23%. Saying all things are equal, and at this point, we really dont know, as Ive said, since AMDs previous coldbug, going to 1.5+ isnt even a 20% increase. If you neglect the starting Voltages, and put the wall at Intels wall, then your right, but again, as Ive said, no one knows. We dont know what limitations HKMG brings also. We dont really know what advantages or diaadvantages SOI vs bulk and vise versa each has. Everyone assumes, or has made mention of past nodes, and peaks.
3.6+ for Intel, and 3.2 for AMD. The problem with that is, Intels process was already tweaked to the max at the time, much more so, than AMDs process, and that also was part of the misconception back when HKMG was first introduced. If you get away from all that, and JUST say everything IS equal, going from 1.35 to a 23% increase would allow for a 1.6+ Voltage, apples to apples. While Im NOT saying this to be the case, as Ive already said we dont really know how thisll go, to make assertions that Intels voltage max has to equal AMDs, or vice versa, just doesnt make sense either

Seriously post the transistor performance information or stfu your rants are getting bothersome and farfetched you have latched onto this just like the "government can solve everything with cash" mantra. You seriously need to do research and stop with the flame bate and the blanket statements. You haven’t even gone into any detail as to how HKMG works yet you proclaim you understand the technology and see fault in it in comparison to bulk SOI which you also have not got into any detail explaining. You’re just another Baron using questionable links to support your weak arguments, while rehashing old topic material that has been gone over with far more detail than you are capable of articulating.

If Jake didn't remove all of Jumping Jacks posts this would have ended a long time ago with you going back to your hole until the next magical fairy inspiration you have.

Word, Playa.

Whoa, not only does JDJ start this fanboy thread, he has well over 50% of the posts in it!

Guess he enjoys talking to himself a wee bit too much...

This one looks like a very fair and open minded thread to me. Can't see what's wrong with it. Jay might show a certain 'green enthusiasm' sometimes, but, after reading him for a while, I guess this all is much more about 'technology' than it is about AMD or Intel.

BTW, damn, I want a Core i7, but now I wanna check Phenom II before going for anything. Problem is I'll feel bad if I go with AM2+ just to see AM3 being released a month after. Then again, I wanted AM3 to have the new SB800, but it looks like it will be released with the SB750... Oh, tech progress.

I agree. He's a technology fanboy!

Saw a post from him over on UAEZone that he gave up on Tom's and even erased the URL from his hotlink toolbar .

Maybe it's JDJ's avatars scaring him off? There has to be a hundred of them in this thread alone

I appreciate that jimmy, I really do. Yes, its possible they tweaked the cpus in that test, but then itd be the most devious thing Ive seen lately, worse than fudging drivers. Not that I really consider it worse, but in AMDs position, itd be far more worse for them.
According to AMD, theyve improved their Branch Prediction, the TLB and shortened the pipeline 7%, which gives a again, according to them, 3% IPC improvement, add in the cache and thats their IPC improvements.

The thermal diode was somehow changed, dont know how, and thats how theyre getting rid of their cold bug. The link I posted to, itll eventually show how people can disable their Th. diode on their i7's, if someones into extreme ocing, so the i7 will get over 5Ghz.

I just think the industry needs this more than anything at this point. If the total impression keeps coming from Intel, and only Intel, stagnation is bound to come, and not for the lack of trying on Intels part, but more in the 2 heads are better than 1 mindset. If this all pans out, and AMD makes even a semi comeback, itll open up new doors of choices, that alone helps prevent stagnation. If its a somewhat different approach to ocing, that also adds to the mix. Ocing and AMD have been out of it, even the K8's werent that great of ocers, especially on a consistant basis, so to have AMD do this now would tuly be phenominal (yes pun intended) and be good for the industry. That would make me happy, not just for myself, but all of us. Im going by more is better, and this would be more