The New AMD Building Thread

wonder how the 780a chipset is

 

http://www.engadget.com/2008/12/23/amd-phenom-ii-processor-bought-benchmarked-coveted/

 

Sam,
That's all well and good, but present Phenom IIs do not have the HKMG Metal Gate Transistors, which will give the Phenom II parity with comparable HKMG Intel Chips. Ati and IBM have jointly developed new low wattage HKMG Transistors for all AMD's 45nm CPUs. The timetable says Mid 09, but I think we will see them sooner than that, rather than later. The performance increase with the addition of HKMG should be about 20% across the board!

AMD was able to get to 45nm without having to use HKMG, and still perform well, while Intel's design Architecture would not allow it. They needed HKMG to make their 45nm chips a reality, while AMD will be adding it to existing 45nm chips. The new low wattage HKMG Transistors will also drop the power consumption about 30-40%. I wouldn't at all be surprised at seeing 7+GHz on LN2 and maybe 4.3 to 4.5GHz on air!

Happy Holidays,
Russ

 

How do you know that? Not saying you're wrong, but what basis is that on? Are they going to release the chips to retail initially without?

 

It's all in some of the articles I posted on this thread. The Current Phenom IIs will be launched Jan. 8, 2009, but will not have the HKMG technology until mid year. I suspect it will be sooner than mid year though. That's just my gut feeling, combined with a bit of inside information. AMD hasn't said very much about it and, they've chosen to let everybody else speak for them! In fact the only denial I've seen AMD make was when the Phenom II Jan. 8 launch was reported to be a paper launch. AMD flatly stated that the product would be available for sale on Jan. 8 and that there would be no paper launch of Phenom II! There's no word yet on whether people buying the early ones will get some sort of upgrade when the HKMG ones hit the stores.

In the past, Intel has poo-pooed every CPU AMD has brought out, but now they haven't said a whole lot about the Phenom IIs at all. An interesting omission!

Best Regards,
Russ

 

Do they need to? I've already seen how the Phenom IIs will perform, and it's nothing to write home about. Maybe I'm blinded by fanboyism, but I've yet to be convinced that HKMG will be the magic bullet AMD need to get them to match Intel's second generation quad cores, let alone their third. Right now everything points to AMD's best phenom being around the level of the Q9300, Intel's one-from-bottom 45nm Quad core. The only advantage AMD will have is if the Phenom II is noticeably less than $270. It'll take a 40% performance boost to get the best Phenom II to match the lowest Core i7 CPU - and I notice the 920 is now priced at a relatively reasonable $300. The only downside is the cost of the rest of the architecture. If Intel pull their finger out and release an affordable LGA1366 chipset, AMD will be left in the dust again.

Don't think for a minute that I don't want AMD to become competitive, it's best for everyone if they do, but I personally think Phenom II is a tiny drop in the ocean compared to the giant gap AMD have to bridge.

 

Sam,
I think you are missing the point. The current Phenom IIs scale well and overclock like crazy. With HKMG they will scale and overclock better and perform better by about 20%. It's not Voodoo or Rocket Science, it's pure math! That bodes very well for a chip that doesn't presently have HKMG, that scales and overclocks well without it. Remember, this is something Intel was not able to accomplish with it's current architecture. They had to have HKMG technology just to get them to 45nm! It was an add-on! AMD left room in their architecture design to allow for future technology, and I think it's going to pay off for them big time!

You mention that if Intel would get their finger out and design a low cost chipset for LGA1333. The problem isn't the chipset, it's all the stops Intel pulled out to get the performance that i7 gives. The motherboards have to support all these little goodies, and they all have to have them, so I don't think there's much chance of cheap motherboards for LGA1333 in the near future! Even if they design an inexpensive LGA1333 MB, it's still going to need expensive DDR3 memory as well as the CPU. Intel still has AMD lost in the dust in terms of speed. but no one is going to spend 2 to 3 times as much money to buy it!

Where all of this matters the most is in the server and workstation markets. If the Phenom II is successful there, Core i7 will be a huge white Elephant for Intel. Without the server market to support it, it simply can't make it as a consumer product, no matter how good it is. With the present economy, it simply costs too much! The cost is much higher for i7 server upgrades as well, what with needing a new CPU, MB and DDR3 memory. Intel has created a huge handicap for themselves, and even if a cheaper MB could be made, it still requires all three items for an upgrade. Point to AMD for maintaining compatibility with current server motherboards and memory. AMD is looking to have 40% of the server market by the end of 2009. From a price performance perspective given the current Economy, they should succeed! Since the Server market is the Bread and Butter for both Intel and AMD, things look very good for AMD at the moment. It's not even necessary for AMD to beat Core i7. They only need to be competitive with the Yorkfields in performance.

Power consumption/performance is going to be the main issue. The power costs to run and cool servers presently cost more for large corporations with large rooms full of servers than the cost of the servers themselves, so they will be having a very critical eye when it comes to power consumption vs the performance they can get. With the new low wattage Ati/IBM designed Metal Gate Transistors, the Phenom IIs should consume 30-40% less than any competitive Intel.

My hat's off to AMD as they did plan well in order to have the right product at the right time for the right price. Some was Dumb luck as well, like the Economy going down the tubes. It hurt Intel and helped AMD. There's no quick fix for the Core architecture either, as it's stretched to the limit right now, while AMD's Barcelona architecture still has plenty of room left for future technological improvements. I've said it before, while Intel may have the better technology, AMD has the better chip architecture at just the right time! They've also got the best designed motherboards to run the Phenom IIs with, in the 780V/700 and 790V/750 series chipsets. It was noted in the Press a few days ago that AMD seemed to be making all the right moves at the right time! I agree! It's also the best opportunity AMD will ever have to gain major market share. I'm pretty sure they know the cost of failure! Failure is not an option!!

Best Regards,
Russ

 

you say that its a bad thing that intel HAD to use HKMG to get to 45nm.... whys is so bad.

great chips with brilliant OCablility

why is it good that oems decide to buy hugely into intel rather than amd?

amd shot themselves in the foot. simply put. its not good that, yes later on in the year they can get better, becuase by that time intel can drop the prices of york fields and i7 alot. they could introduce the P55 early, with new proccessors which can use DDR2.

people within a few months can go out and buy an x48 or p45 and a Q9550 or Q6600 and get great OCs and it will be more than enough for alot of things.

or people with those sytems already who want an upgrade will go to i7 as it is, because they want more power, and in the end DDR3 will be needed.

for AMD DDR3, you will need to buy a new mobo and DDR3 aswell..... huh....

and umm speaking of upgrading costs, for anyone who does wait, alot are intel users with P4/C2D (who would pop in a 45nm) or go AMD DDR3 and new mobo and RAM or DDR2 with new mobo.

even AM2 users, unless they have the good 790GX chipset and want to OC will need to upgrade their mobo.

until do get of their arses, why give praise on false hope?

i am wanting the GTX 295, and hopeing it will be good so i can upgrade, but saying facts without backing it up is BS.

i do hope AMD get back in the game, as ATI have done, and then nvidia hopefully do, as for the consumer its great, but untill they do, dont go with them

 

I'll leave you with this question. Don't you think that the 780G chipset came out when it did for a reason? I do! Without it and it's newer siblings Phenom II has no platform! That's why. It looks very much to me like AMD has had a plan for a while now!

Happy Holidays,
Russ

 

i think it came out for the low powered x2 CPUs they released to capture the HTPC and low end oem market, which they did .

sorry for the AM2 chipset, i meant both AM2 and + because im sure alot of AM2 users will be upgrading from K8 x2s and they will want to OC, but without an AM2+ mobo it wont be too much of an OC.

and as to i7 and DDR3, you could go and get a P45 or X48 and DDR3, also after seeing lots of benches, latency is not much of an issue, cas 9 is fine, hell even 1333MHz DDR3 with cas9 will be fine, its not like the performance drop is significant if there, vs 1333MHz cas 7.

the thing is though, why have a plan of failing for a while and then bringing out something big?

 

I think AMD got caught with it's pants down with Core 2. I don't think they had a clue as to how good they were going to be. I don't think AMD planned to fail, it's just they had no choice at the time. They couldn't very well just quit! Remember too, successes and Failures seem to come in bunches. AMD didn't seem to have very much in the way of success until after the Barcelona. Even then things were very luke-warm for quite a while. Let's face it, the Phenoms were nothing to brag about, yet the Phenom II is built on the same architecture, but reduced to 45nm and is much faster, more powerful and much more overclockable than any AMD chip ever seen before, and it's going to bet even better. The big question is when? My sources tell me that HKMG will be early. Wouldn't it be something if they were ready, just waiting for Jan. 8. That would be something to see! Still, the facts point to something just like that to happen. I sure don't think it was an accident that the 780G chipset gave socket AM2 owners a 15 to 20% boost in performance, which strangely enough gave the Phenom II about the same boost over previous Phenoms. Another 20% from HKMG and we are right at that 40% Sam mentioned. Another plus is that the 780/790 chipsets will have some time for the bios to mature so that when HKMG does hit, the MBs will be ready and seasoned for them.

You said:"why have a plan of failing for a while and then bringing out something big"? My best guess is with what I've been seeing for months with all these seemingly little things coming together the way they have. The chipsets and the instant performance boost for AM2 owners, the current Phenom II all of a sudden making a big splash, being able to natively use both DDR2 and DDR3, AM3 being backward compatible to AM2+ and even some AM2s. All of this backwards compatibility leaves a large choice of combination's of performance levels available to the customers, mostly in the Server Market. With hundreds of millions of dollars at stake or more, I think AMD is in the best position possible to win that market, and I think they are going to try to do just that! Core i7 while fast, just isn't cost effective with the horrible economy we have now. The Yorkfields can compete for speed, but they won't be able to compete in terms of watts consumed for very long. For once the old joke about AMD and NegaWatts isn't such a joke anymore! There the big edge has to go to AMD as the MKMG Transistors consume 30-40% less power than the ones Intel uses in it's chips, and they are an AMD exclusive, developed by Ati/IBM. If somebody wants them they will have to develop their own or hire someone to do it for them from scratch, as they have both US and International Patents. It just seems to me that the timing would be perfect if the "Real" Phenom II stood up on Jan. 8! That's how I would do it if I ran AMD, and Jan. 8 would have been my target to drop the big bomb. The thought of the possibility of that happening has been nagging at me for a month now! Must be my "Devious Mind" at work again. Still, if that was to happen then the Server chips would also have to come out the same day as well! That would be Da Bomb to end all Bombs! Gotta think that one over! LOL!!

On the latency, it's still lost clock cycles. If I was going to build one it would be Cas 6. No point in spending that kind of money and not put in the very best! Not when I can get 6GB (3x2GB) of Cas 6 running at 6-6-6-24 for under $300. The difference between Cas 9 and Cas 6 is very noticible, very much like moving up to the Cas 5 1066 Dominators from the Cas 4 800MHz G.Skill I had before in appearence. A much fatter memory bandwidth too! Even some overclocking with looser timings. I'm supposed to build two Core i7 965s in the very near future, and she wants the very best cutting edge goodies. I'm just waiting for the money to get here. She's an old customer I've taken care of for years, so it shouldn't be too long!

Happy Holidays,
Russ

 

mmm, i dont think i had ANY difference moving from 555 15 800 to 555 15 1066....

bar the tiny difference in benchmarks....lool

why not release the phantom II with the HKMG?

 

Supposedly they are not ready yet, still in testing or some garbage like that. maybe attract more interest! I don't know! Jan. 8, 2009 would be good for me! ROFLMAO!!

Russ

 

Russ


You guys really have to shorten your posts a bit or no one new coming to this thread is going to make an attempt to follow it.

I've learned to never count either chipset manufacturer out. When one finds a technology that surpasses the other, the other doesn't just give up they head for the drawing board.

AMD ran Intel into the ground for several years before Intel finall saw a need to catch up. What took Intel so long to do it? Simple! Intel controlled the market by the use of unfair business practices to prevent AMD from entering into certain markets. As long as Intel had PC manufacturers such as Dell in their pocket they had no need to improve the quality of their processors. Once companies like Dell began building AMD rigs because of their processors superior processor performance Intel awakened.

Much of the i7's architecture was borrowed from ideas taken from AMD. The on onboard memory controller is a great example which eliminated the FSB. AMD also developed the first true quad core well in advance of the i7 while Intel stacked two dual cores on a single die to get a quad core processor to market. Remember everyone figured that Nvidia had finally buried ATI and then along came the 4000 series which didn't attempt to take on Nvidia's GTX280. ATI chose instead to blow away Nvidia's low high end and down GPUs which is where most of the market is.

Right now AMD is still doing quite well in the server market because of its architecture and will continue to do so until i7 finally unseats it. By then AMD could have a new offering that will compete with the i7. Rumor mills suggests that AMD has an ace up their sleeves. I say it's time for a little competition to get things moving again.

http://it.anandtech.com/IT/showdoc.aspx?i=3484&p=13

 

3Ghz to 3.85Ghz is better than AMD's previous efforts, but hardly 'like crazy'. Intel's Core 2 Quads have managed 2.66 to 3.85. You will notice that while HKMG may have allowed Intel to run at 45nm architecture, it did not mean their chips overclocked any better than before. The 2.4Ghz 65nm Core 2 Quads overclocked to 3.8Ghz on their best stepping, nearly 60%, AMD haven't even reached 30%. You could argue that at least AMD have reached the same clock speed as Intel, but that's not really relevant when a 3Ghz AMD is beaten by a 2.5Ghz Intel.
As much as you're certain HKMG is going to give AMD a vast performance gain, I just don't see it, I think there's more to Intel's die-shrink than HKMG alone, and the whole orchestra gave them barely 10% higher performance per mhz, and a 10% higher initial clock speed, with no increases in % overclock, decreases in fact.
i7, certainly initially seemed a waste, and to gamers it still does. To video encoders though, the i7 920 is $300, you don't need to spend megabucks to buy i7 boards and RAM these days, a lot more than 775 I grant you, but considering you're getting pretty much double the performance of AMD's best Phenom II, that's worth paying for.

Oh come on now, straight from the mouth of AMD. More faster and More powerful not meaning the same thing? 'much faster?' The 9850 barely matched the Q6600, the 9950 barely beat it, the Phenom II 940 is still going to barely beat it.

- Pentium D and Core 2 Duo anyone?

The i7 is a somewhat intelligent design step, but half of it is just strapping all the addons they needed to get P4 to perform properly to a Quad core architecture. AMD's genuine quad core approach was far more sophisticated from a design perspective, but just shoving two Core 2 Duos on the same chip seemed to work pretty well anyway.

 

sammorris,

It matters not what the 9850 can do as the reference was to the coming 45nm Phenom II. It was also a comparison of Phenom II to existing AMD processors, not Intel. It says "much more overclockable than any AMD chip ever seen"!

HKMG was also a " necessary work around" for Intel, and not even part of the original design, and as you pointed out, didn't do a lot for the performance. To me it was a little like putting the supercharger in the trunk rather than under the hood! It's remarkable that the Phenom II does as well as it does without HKMG. Remember also that the "almost 4GHz" was done without a particularly decent cooler, barely more than the stock heatpipe cooler that comes with it. HKMG properly implemented into the design should yield about a 20% improvement in the amount of work the processor should be able to do for any given clock speed.

Russ

 

Hmm, I don't think 4Ghz on air is passable, for the main reason that AMD could only get 4.4Ghz on Phase.

 

Sam

Shouldn't you be in bed for fear that Santa will see you up and leave you pickles?

What the heck are you guys debating over? Both have their merits.


Merry Christmas Sam although I know that the hour has already passed and merry Christmas Russ. Now stop this you are boring me to death!!!

 

Actually no, 58 minutes to go

It might seem petty, but Russ still seems to think Phenom II will be the best thing since sliced bread. I'm just not quite that optimistic

 

58 minutes isn't a long time to go into a coma for fear of surprising Santa!

Nothing is better than sliced bread, just ask the Earl of Sandwich!

Both formats have unique attributes and sometimes one rises to the top like butter in cream. I can show you posts where you and I both were on the AMD wagon going west. It's sensible to be fickle when it comes to a brand of technology, we buy them, not marry them.LOL