Hope for AMD?

[turpit]

amd-to-intro-45nm-quad-core-phenom-this-quarter

 

[4745454b]

Unless 45nm allows them to increase the clock speed, it won't make much difference. More yields per wafer, but I'm not sure that will be enough. The best hope for AMD until the next arch, is for their video cards to get back on top. 4870 might do it, but not enough is known yet to say for sure.

 

[yipsl]

Tom's reports (from Digitimes) that the first two Deneb's introduced this year will only clock at 2.4-2.7 and 2.5-2.8. The word's from motherboard manufacturer's, so I guess AMD's still working on getting the stock clocks higher. The 9750's and 9850's will be moved to 95 watt versions as well.

So, slim hope, but it's not the godsend that many were thinking 45nm would be. AMD will be in a bind with clockspeed, and in a budget CPU market until Bulldozer, which AMD promises will leave K8 and K10 behind.

Personally, I hope their promises pan out this time. AMD was complacent with K8, and not only was K10 late, it wasn't much more than a 10% per core improvement. That improvement was lost because clock speeds couldn't match the fastest 125 watt X2's.

 

[4745454b]

I'm not impressed with bulldozer however, at least with what little we know. K10 was supposed to be the god sent CPU that AMD was hoping for. As you correctly pointed out, the improvements per core was a big "who the F cares", seeing as the clocks went down.

If by "radically different" AMD means it will have a GPU as one of the cores, or has a GPU integrated into the CPU, then I doubt that will increase performance as much as some think. AMD needs a CPU that decrease thermals/power consumption. (125W and soon to arrive 140W phenoms? Give us a break!) They need a CPU that can get beyond 3GHz. (C2Ds can overclock beyond 4GHz, and they are faster to) Adding the equivalent of a 3870 isn't going to help with thermals. It makes no sence to add a low end GPU, as thats what onboard video is for.

 

[smalltime0]

I personally think that a no screw-up falcon release with be the save for AMD... As long as it works well it should steam-roll the competition (then again nothing is currently going AMD's way (even though I own a Phenom and a X2) so..)

 

[yomamafor1]

Personally I have my doubts in this new's accuracy. If I remembered correctly, AMD still couldn't churn out a 45nm prototype until Jan of 08. Add in the time to tune the process, ramp the process, and iron out the bugs, mass production, it will be at least Q4 of 08 before they actually launch the product in any significant volume.

But heck, I'm just a brainless Intel fanboy. I guess we'll have to wait and see.

 

[boonality]

I have always gone back and forth between intel and AMD. Completely by coincidence too.

Intel inside 25MHz
AMD K6-2 333MHz
Intel P4 2.4GHz
AMD 3700+
Intel C2Q

Next one for me might as well be an AMD if all holds true. But seriously though, AMD is great for people in a tight budget, you can get a 5000+ X2 black for $85!

 

[Iain1974]

So does this mean Deneb will debut at 2.4 and 2.5GHz? I'm assuming the we should realistically look for the lower end of projections? And perhaps 5-10% improvement? Lowe power is a good thing. Overclockability isn't something we'll know about until November/December.

Something to cling to I suppose, but not much.

 

[Amiga500]

amd-to-intro-45nm-quad-core-phenom-this-quarter

Doubt it will help the desktop market much, as there still seems to be a clock ceiling well under 3 GHz.


The better thermals should be a good plus for the server market with Shanghai (as will the added L3 cache and the 6 core Istanbul).

I'm not impressed with bulldozer however, at least with what little we know.

If by "radically different" AMD means it will have a GPU as one of the cores, or has a GPU integrated into the CPU, then I doubt that will increase performance as much as some think.

How can you possibly come to any conclusion regarding Bulldozer already?!?!



Do you not realise the advantages of a high number of parallel (but limited) instruction sets? Some media encoding, workstation work, HPC work, some server apps - will all benefit tremendously.

For example, you have a 8 core CPU, but within that, you can do a 32 thread encode.

 

[Reynod]

All my optimism evaporated after the disappointment of RV600 followed by Barcelona / Phenom.

The hype for both was just too much to bear ... for a fanboi ... Iv'e become disillusioned ... despondent ... that's just stupid.

It seems like they come up with great ideas but can't produce the final goods - both technologies look good on paper but in the real world only seem to compete in only a very narrow range ... ATI's chips poor general performance despite the massive memory bus ... Phenom's IPC despite the IMC advantage ... and low clocks.

Yeas I realise Barcelona is primarily designed for multisocket .... but even then it only seems to have an advantage on 4 and up ... which has largely been lost of recent times with the newer Zeon's being quite impressive ... and cooler too.

I think it is clear they threw their advantage away and surfed the Opteron board into the rocks ... that's poor management.

They clearly can't pull any rabbits out of the hat quickly either.

Note Ed at Overclockers has a great little article about that missing dual core ... Puma .... there ain't one ... they didn't do a mask. Now they don't have any cash to do one quickly.

While Intel's approach has been to build on its success and fine tune every possible avenue (cache / prefetch logic) AMD has wasted a lot of time getting 4 cores from previous generation technology to talk to each other on a piece of silicon too large to operationalize on an outdated process.

I bet at least half the Engineers there must have realised this ... but shutup for fear of their jobs.

It's like they designed a souped up pickup and then entered it in a rally race.

The circuit changed ... they forgot.

The core 2 architecture bolted to an OMC I am sure will be impressive.

If AMD does have an answer it had better be Pure Pwnage !!

But I truly doubt it will be.

I apologise to the AMD fanbois out there ... Iv'e lost my Mojo.

 

[keithlm]

I sit back and watch and I find great humor in this subject.

I've seen the dogmatic responses in these forums go from: "Oh AMD won't have anything in 45nm until 2009; they just stink."

Morph into: "Oh it will be 4Q2008... they are just worthless."

To the latest dogmatic claim: "They're releasing 45nm this quarter? It won't matter. They'll be a disappointment. They won't be very fast." (NOTE: Just because the other company's 45nm offerings were a bit underwhelming doesn't mean anything. Other than many people were disappointed.)

If I remember correctly, the posts (on this forum) all said we won't see a B3 stepping until 3Q2008. I'm waiting... waiting... oh WAIT... it's already IN MY MACHINE. (Now if I can get Gigabyte to fix some BIOS issues... that would be nice. But I'll probably ditch them and go DFI. It would be worth the added money to get rid of the headache.)

Having released the B3's... if they also get some 45nm released THIS QUARTER... then that IS "pulling a rabbit from a hat". Regardless of any bias... people should still be able to give credit where it is due. (I know.. that's kind of like asking people to actually use and understand LOGIC.)

 

[maximiza]

All the execs that have been jumping AMDs ship and the cash problem means they are going to have to get bailed out. I im thinking the German goverment might do it to keep the Fab in Dresden open. They went through all their credit and need cash from someone.

 

[jimmysmitty]

Well its possible for them to release a small quantity of this chip this quarter but probably very late.

I don't see a major performance increase with the die shrink as even AMD has had the same results as Intel with that part. Normally its lower heat and power usage .

As I said before unless by something short of a miracle its just a die shrink and we can hope it does lower thermals and works right.

But I am still waiting to see it compete with Penryn and then Nehalem.

 

[JAYDEEJOHN]

Everyone complains about mismanagement, and then they say "see all the management bailing" and at the same time AMD itself says theyre getting rid of the fat, the non producers etc. Sounds to me theyre doing exactly what they need to do. Having a gun pointed at your head for 2 years has an effect. But saying when are they gonna crumble, and yet theyre still standing does merit some respect. Were all hoping for a good thing from them. The 4870s will be a nice addition, timely as well. I agree, the recomendation here is get a Q6600 instead of any of the 45nm's, so obviously, no ones that impressed with Intels new offerings. Maybe a time for AMD to tweak a little closer, clock a little higher, pull that gap in tighter, and who knows? If theyre still standing, they may yet surprise us

 

[someguy7]

Will you people please just stop drinking the AMD kool aid. This is just getting ridiculous.

"(NOTE: Just because the other company's 45nm offerings were a bit underwhelming doesn't mean anything. Other than many people were disappointed.)" What are you smoking? And this one "I agree, the recomendation here is get a Q6600 instead of any of the 45nm's, so obviously, no ones that impressed with Intels new offerings."

Lets start with people recommending the q6600 over its 45nm counterpart the q9300. Lets start with the obvious. Its cheaper and just had a price cut. It also gets recommending for people who overclock, escpially on cheaper/older mobos cause of its higher multi/lower fsb. It also happens that most of the folks on hardware forums do overclock. This is the best part here. The fanboys use the logic that the majority of pc users will never overclock(which is true) when they support the companys chips that can barely get a extra 300mhz out of them. But then when Intels replacemnt for the q6600 comes out isnt as overclocking friendly as its older sibling now nobody is impressed, or they are a bit underwhelming. So now the logic is flipped upside down. The new cpus still do overclock just not as high and easy as the older one. They are faster clock for clock. They use less power/heat. But they are not impressive because they cant clock as high the q6600. You guys would be having a party if the Unlocked black edition b3 could overclock as high as non overclock friendly q9300.

Then you got the entire wolfdale line. They are underwhelming? Ha. They are exactly what they where supposed to be. A die shrink with some minor performance improvements. Use less power and create less heat. Oh and yeah they overlock like crazy.

Got the nervs to say use logic. You guys are just as logical as thunderman/baron. Keep on drinking the AMD kool aid.

 

[EXT64]

Well, the glimmer of hope is that Intel said that AMD was crazy for trying that design in 65nm and they themselves are about to do it in 45nm. I guess time will tell, for now I'll just stay neutral.

 

[JAYDEEJOHN]

Well, since Im a you guys guy (whatever that means) do a check on the posts. Its true. Did I say anything in regards to anything other than it was a disappointment? Glad you said all that about Wolfdales, as its all true. But still, I read what I read. Check the posts. Why does this upset you so much? Im not spreading FUD, just looking at the responses , the posts, the recomendations. I cant believe that what Ive said isnt so, but Im innocent until proven guilty. Tell me whats been recomended here more than anything else? Wolfdale? (since its been out) or the older 65nm? Let me see, a brand new line comes out, but its not recomended? OK, Ill give you a cost effect, but we are enthusiasts arent we? Therefore? We want the most we can get, bang for the buck sometimes, but sometimes just the best. OK, going by your logic, the best is the oldest not the newest. OCs better, cheaper. Now lets go to the newest. It takes less power. How enthusiastic is that? I see a slight let down for us enthusiasts when it comes to Wolfdale, I see the recomendations for the 6600 and thats just the way it is

 

[JAYDEEJOHN]

Listen, just because I typed some hope for AMD, Im not alone in that. Im not a fanboy, far from it. I was looking forward to 4+Ghz Wolfys on air too. Dont attack my words when at least Im disappointed with the Woldales. I dont have a side, so say what you want about AMD, that doesnt make me less diappointed by the 45nm's. Im hoping that AMD's 45nm's will do better than their 65's, as theyll need alot to catch up to Intel at 45

 

[someguy7]

The wolfdales ARE recommended over the 65nms. Who is recommending a e6850 over a e8400? A e6750 over the E8200. There is no downside(except for the temperature reading issue) for the enthusiast with the 45nm wolfdales. They are better in every single way. Yeah they use less power and ARE FASTER then the older chips. They are faster clock for clock, OC higher because they run cooler. The only time you will see anyone say get the 65nm over the 45nm is when there is no comparable chip. Like a low budget build. There is no 45nm equivalant of a E2160. So if your on a tight buget and are a overlcocker that makes the most sense.

You do not see any recommendations for the 6600 as you put it. You go on your little rant about the wolfdales then you use the q6600(i assume) as your point. The wolfdales are DUAL CORE not quads. The e8400 is also right around the same price as the q6600 right now. So if somebody is building a machine and using programs that will take advantage of the quad, anybody with a brain will say get the quad. If you somehow think that is a disapointment then your logic is flawed

 

[JAYDEEJOHN]

Im not talking about the wolfys, as they do oc over 4 Ghz, or up to. My point here was about the quads. Im sorry if I screwed the hell out of that post. The wolfys, as you and I both said, are great, its the quads that are the disappointments. They dont oc as high as was hoped.

 

[JAYDEEJOHN]

I defined the terms of my disappointment. The quads dont oc as well as I hoped, and Im not alone. Im not saying they arent good, or dont do as they claim to do. But youve been here long ehough to know what was expected/hoped out of these quads. It came in short. Is that the end of the world? No, but to me a disappointment nonetheless

 

[keithlm]

Got the nervs to say use logic. You guys are just as logical as thunderman/baron. Keep on drinking the AMD kool aid.

The fact is that the 45nm chips are a disappointment to many people. Especially the ones that were expecting gains in top overclock speed.

You can try to justify the situation all you like. It won't change that simple fact.

In fact: Please do keep trying. It is amusing. Bereft of facts AND logic... but amusing.

 

[yomamafor1]

I defined the terms of my disappointment. The quads dont oc as well as I hoped, and Im not alone. Im not saying they arent good, or dont do as they claim to do. But youve been here long ehough to know what was expected/hoped out of these quads. It came in short. Is that the end of the world? No, but to me a disappointment nonetheless

I completely agree with you. Intel's FSB 1333 CPUs are disappointments in general, especially the quads.

However if you really think about it, why would Intel gave you the capability of buying an entry level quad, and OC it beyond its flag ship product, completely free? It takes a big chunk of high margin of selling high end quads out. If I were Paul Otellini, I would probably do the same thing too.

But again, as end users, not being able to OC do suck.

 

[JAYDEEJOHN]

My question then is, are the highend really that good? Anyone with some benches? Maybe, as you said, if theyre holding them back, the only way (most likely) would be binning. Then the question comes, is it worth it? 1100 or more USD for one. Are the top that much better?

 

[someguy7]

Keith what are you talking about now. Everybody knew from the moment all the details came out about the new quads(q9300) what there over clocking potential was. As soon is people knew what the multi was going to be, anybody with any overlocking know how knew right away they were not going to be the best overlockers due to FSB limits.

The chips are only a let down to those people who where looking for a higher top overclocked speed on the lowest end of the new cpus.

To bad AMDs overclocking friendly quad core doesnt even overclock as good as Intels worst overclocking quad core. But its a success because it does overlock better than the overlocking/clockspeed disaster that it replaced. Dont worry though, somehow when its shrunk to 45nm its going to turn into a beast. HAHA.