AMD AM2 To arrive early! + New intel cores

[BeepBeep]

Just caught wind of the early release of the AM2. AMD are *supposedly* releasing AM2 early to sidestep Intel’s counter release of a new core on the original AM2 release date.

Nice to see a fight in which we win!

AMD's AM2 Launch moved Up

 

[Dunky]

Releasing the product early doesn't really do AMD any favours. It doesn't really counter the fact that Conroe is ahead in benchmarks. Intel just now simply have the superior product and AMD's main audience (enthusiasts) will simply wait a bit longer for Conroe. Although Intel could do the same (possible but probably not likely) and bring the date forward - actually Intel are probably in the better position todo so.

 

[HeX]

AMD could bring it out next week and I still wouldn't buy it taking that it offers no performance benefit over 939 and the fact Conroe kicks it squar in the nuts in performace.

I'll hold out til Conroe thanks.

 

[ihatelag]

and the fact Conroe kicks it squar in the nuts in performace.

I'll hold out til Conroe thanks.

Hehe are you sure you're not being a bit hasty with your assessment...the conroe processor hasn't even been released...imagine if Intel takes a page out of Sony's book and decides to delay the release for a further 8 months or longer...AMD may come out with a superior architecture compared to conroe by then.

I reserve judgement on who has the best processor until the actual conroe processor is released. Also the conroe processor has only been seen in test enviroments..I would like to see real tests made on it.

 

[Dunky]

Hehe are you sure you're not being a bit hasty with your assessment...the conroe processor hasn't even been released...imagine if Intel takes a page out of Sony's book and decides to delay the release for a further 8 months or longer...AMD may come out with a superior architecture compared to conroe by then.

I reserve judgement on who has the best processor until the actual conroe processor is released. Also the conroe processor has only been seen in test enviroments..I would like to see real tests made on it.

Nah they won't delay it - they've got decent enough engineering samples out now. If they want the market (which is swaying towards them with Conroe) then they won't delay unless it's a major showstopper. Even then, most people are happy with their 939 setups so might be willing to wait if it is delayed.

The Anandtech review on Conroe was pretty well done, there were a few discrepancies in the first test so Anandtech re-ran the tests with changes (suggested via comments to the review) and the Conroe still outperformed. Also if you have a look over at XS, some people there have ES of Conroe and their tests are pretty consistant with the Anandtech's review.

Reviews usually have a mix of synthetic (3DMark, Aquamark) and real (Video Encoding, Gaming) but the thing is most review sites use roughly the same benchmarks hence even with synthetic benchmarks - it's still an even comparison.

 

[HeX]

Nah they won't delay it - they've got decent enough engineering samples out now. If they want the market (which is swaying towards them with Conroe) then they won't delay unless it's a major showstopper. Even then, most people are happy with their 939 setups so might be willing to wait if it is delayed.

The Anandtech review on Conroe was pretty well done, there were a few discrepancies in the first test so Anandtech re-ran the tests with changes (suggested via comments to the review) and the Conroe still outperformed. Also if you have a look over at XS, some people there have ES of Conroe and their tests are pretty consistant with the Anandtech's review.

Reviews usually have a mix of synthetic (3DMark, Aquamark) and real (Video Encoding, Gaming) but the thing is most review sites use roughly the same benchmarks hence even with synthetic benchmarks - it's still an even comparison.

Indeed, in every test of the ES Conroes i've seen the 2.63 Conroe has comfortably beaten the FX62 (they oc'd a FX60 @ 2.8Ghz).

The true AMD competetor to Conroe will be the K8L but thats not sceduald for release until Q2 07 a good 5-7months after Conroe is due to hit the shelves.

I wont go AM2 for the same reason i've still not gone 939, the performance difference between the sockets and chips is still relitivley small. I admit AM2 is a bigger step forward than 939 was adding DDR2 and next gen chip support, but in its current incarnation which consist of standard A64 cores with DDR2-800 mem controlers I can live without and will hapily wait for Conroe to arrive and give me a good speedboost over my current setup.

 

[NathanE]

I thought K8L was the DDR2 chips...?

Considering some new K10 information was leaked a couple days ago (Hyperthreading v2.0) I'd guess that it is closer to production than AMD has been letting on. IMO, K10 is the true competitor to Intel's Core architecture. Although Core is currently crippled due to it using the old FSB design. As soon as CSI is finished Core will scale very comfortably up to quad-core. At the moment it is really limited to dual core for mainstream desktop chips.

 

[warnea1984]

I've seen a few ES core benchmarks for the intels on certain forums as well but to be honest you can't even trust them, keep in mind that major players in the electronics industry realise the selling potential of hype, and actively post comments on forums to create hype around there products, by telling everyone how they have samples of the cores and how good they are and that when there released we should all blindly trust them and buy the product. I am not saying that the results are false its just best to wait until there avaliable to consumers and then wait for reviews and comments from the consumer products not from ES.

Remember TRUST NO ONE!

 

[Pulseammo]

I thought K8L was the DDR2 chips...?

mmm, dont think so. I think K8L are going to be AMD's first 65nm chips if i remember rightly?

 

[Kamakazie!]

K8L are the first 65nm chips and are arriving Q1 07 from what i have read. Not Q2. K8L is gonna be a serious competitor to Conroe... as for K10, i am not really sure what this is meant to be, i understand its the next new architecture but all i have seen about it is that AMD are working on "anti-hyperthreading" for it.
Im gonna go AM2 almost certainly mainly for the upgrade path its gonna give what with K8L and later revisions meant to be supported by it.

 

[Dunky]

I've seen a few ES core benchmarks for the intels on certain forums as well but to be honest you can't even trust them, keep in mind that major players in the electronics industry realise the selling potential of hype, and actively post comments on forums to create hype around there products, by telling everyone how they have samples of the cores and how good they are and that when there released we should all blindly trust them and buy the product. I am not saying that the results are false its just best to wait until there avaliable to consumers and then wait for reviews and comments from the consumer products not from ES.

Remember TRUST NO ONE!

It's true that certain ES samples are hand picked. But the guys at XS have been pretty consistant with their testing and the final product. At the end of the day however, normal consumers don't overclock like the guys at XS. XS guys have also probably played with nearly every CPU on the market, so they are able to give a better comparison between different CPU's - something a consumer probably wouldn't be able to do. Of course since these are ES, it should be taken with a bit of salt (this includes AM2) as there will be no doubt some tweaks made to the processor before it hits full production.

 

[LordByron]

Im gonna go AM2 almost certainly mainly for the upgrade path its gonna give what with K8L and later revisions meant to be supported by it.

Who's to say that Intel have not got a good path of upgrade with the new Coroe setup? It might be foolish to assume that Intel have not got anything planned?

 

[Dunky]

Im gonna go AM2 almost certainly mainly for the upgrade path its gonna give what with K8L and later revisions meant to be supported by it.

K8L really doesn't bring anything completely new that Intel can't counter. Intel are already on 65nm, they'll probably bring out a quad-core at some point (probably around Q1 2007), DDR3 is pin compatibile with DDR2 although latencies are still mahoosive for DDR3 which does AMD no favours. The only thing which it might bring that will be of interest is this talk of multiple FPUs. Fine this might improve floating point calculations but not everything uses FP calculations so don't expect a total increase across the board. Also there's talk of K8L moving onto a different socket (AM3) but I'd take this with a bag of salt at this time. It's too early to tell. But we can take a good guess that K10 will probably require another socket.

Until we know exactly what the future of AM2 is, it would be unwise to jump for the sake of K8L.

 

[Corasik]

Latest news I read today is that Intel are also preparing to lanch Conroe early, Originally targetted for Q4( November) release, Intel are now intending to launch them in Q3, so it could be as soon as June. The 'paper' launch of conroe should be made in May.

Sure AMD may have new products in the works, but Conroe is 'possibly' out performing the AMD64's while still using a legacy FSB. If AMD come up with something better, Intel can develop an integraded memory controller, or add systems like Hyperchannel and introduce it as the 'next' model.

P4 was not a good enough processor to bother developing a better bus, but if conroe is bus limited, and AMD throw out a performance challenge, there is nothing to stop intel improving on Conroe.

 

[Dunky]

Sure AMD may have new products in the works, but Conroe is 'possibly' out performing the AMD64's while still using a legacy FSB. If AMD come up with something better, Intel can develop an integraded memory controller, or add systems like Hyperchannel and introduce it as the 'next' model.

P4 was not a good enough processor to bother developing a better bus, but if conroe is bus limited, and AMD throw out a performance challenge, there is nothing to stop intel improving on Conroe.

Which is exactly what they are working on just now with CSI (equiv of HyperTransport). Processors for use with CSI will have integrated memory controllers and will probably use Conroe type procs. Looking at 2008 for this though however a few people round here think it will probably be Q4 of 2007.

 

[Biggles 266]

The AM2 chips are just normal K8s. They have a different memory controller because of DDR2 and they'll be running at faster speeds, but only a couple of notches higher. I think the 5000 X2 and 5200 X2 for instance are scheduled to be AM2.

The K8L is a while away yet. It's going to have phenomenal FPU performance, better than that of Conroe/Merom/Woodcrest. It's not something that will make a huge difference to most people, but it will on certain types of server and workstation uses. That, coupled with the fact that the Intel Core architecture still will be using an old style FSB, means that AMD will still be considerably better performing for anything above 2 sockets. Quad-core Core might saturate the FSB as well, meaning that they aren't all they're cracked up to be. Dual-FSB will partially solve this problem, but it will take CSI to solve this. And I don't see it being early.

I'm not denying that the Core architecture is better than the K8. It should be, considering that the K8 is 3 years old. The only thing I'm disputing is the gap in performance that will exist. I've seen a few reports, to be taken with a pinch of salt, that indicate that Core isn't so hot at multithreaded stuff and that quite a few of the benchmarks are synthetically high because of the large cache.

The K10 is what I'm really looking forward too. If, and it's a big if, AMD manage to get the reverse Hyper-Threading thing working well, then single threaded performance will shoot through the roof. For those that aren't aware of what it is, it's a way of executing a single thread over multiple cores. By then AMD will be at 65 nm, maybe even 45 nm, so I'd be expecting clock speeds of perhaps 4 GHz if there isn't a radical departure in design.

Still, there is life left in the K8 yet. Once at 65 nm, we can expect it to ramp up again in terms of clock speed. AMD did manage to produce 3 GHz Opterons on 90 nm, so 3.8-4 GHz K8s is a possibility. That'd be outperforming a 3 GHz Conroe. It's incredibly unlikely however that they would increase the cache size, it'd hurt latencies too much.

The way I see it, sure Conroe is going to be king of the hill upon release. Just like the Athlon 64 was upon its release. But AMD aren't mugs, they've had years to work on something. Just because we don't know entirely what it is, doesn't mean it won't happen. Just like the last days of the Athlon XP line when the P4Cs with 800 MHz FSBs and HT were the fastest, AMD will be behind Intel. But they'll be close enough in performance that few regular Joes will notice the difference.

And the best thing of all, to remain competitive in terms of sales, if you can't beat them on performance, you have to beat them on price. Which is why I predict that AMD will slash the price of the Athlon 64s on arrival of Conroe.

At the end of the day, we are overclockers. Benchmarks at stock are pointless since most of us will be running far above that. We're going to hit the jackpot either way, either we get all-conquering performance, or we get great price/performance.

 

[Dunky]

The AM2 chips are just normal K8s. They have a different memory controller because of DDR2 and they'll be running at faster speeds, but only a couple of notches higher. I think the 5000 X2 and 5200 X2 for instance are scheduled to be AM2.

Yep true, as far as people are concerned, it's not really an upgrade to AM2 more of a re-grade.

That, coupled with the fact that the Intel Core architecture still will be using an old style FSB, means that AMD will still be considerably better performing for anything above 2 sockets. Quad-core Core might saturate the FSB as well, meaning that they aren't all they're cracked up to be. Dual-FSB will partially solve this problem, but it will take CSI to solve this. And I don't see it being early.

AMD will still have the server/Opteron market and this is where we'll probably see Quad-Cores really take off - they may have some impact in the workstation market but I doubt they'll dent the desktop market. Home users don't really need a Quad-Core, dual would do for multi-tasking. Unless more desktop applications scaled well, then I don't think Quads will marked at the consumer. Also with 4 cores, I doubt it will appeal to the overclockers (that's four cores to prime - many people find that certain dual cores have one better than the other...this would be a lot worse with x4 cores).

I'm not denying that the Core architecture is better than the K8. It should be, considering that the K8 is 3 years old.

Also bare in mind how old the FSB/seperate mem controller system that Conroe still uses. The fact that Intel is able to get such good performance out of an almost legacy FSB type architecture is feat in itself. Ok it doesn't scale too well to 4 cores but we want to compare apples with apples (dual core).

The K10 is what I'm really looking forward too. If, and it's a big if, AMD manage to get the reverse Hyper-Threading thing working well, then single threaded performance will shoot through the roof. For those that aren't aware of what it is, it's a way of executing a single thread over multiple cores.

I feel this is more of marketing gimmick by AMD. There's really no use for such a technology except as a proof of concept. This type of technology will probably give developers an excuse to not producing fully multi-threaded and scalable programs - It's basically a step backwards. We should be developing programs which use threads so that programs will scale well. Also such a technology will probably not scale well as it would only be able to split a thread up so much. At the end of the day, there are very few cpu-intensive applications that are still single-threaded. Games are one of them however this is changing with multi-threaded drivers and games starting to appear. Music and Video compression algorithms are largely multi-threaded so are large applications like Photoshop. In conclusion - I wouldn't get too excited about it.

By then AMD will be at 65 nm, maybe even 45 nm, so I'd be expecting clock speeds of perhaps 4 GHz if there isn't a radical departure in design.

Intel are looking to start shipping 45nm in 2008 - although I think we'll see CSI on 65nm, I think it will come to life fully on 45nm. AMD probably won't hit 45nm til 2009/2010. But I don't know if we'll hit the 4Ghz barrier with AMD. I think they'll concentrate on additional cores. But we shall see..

 

[Vegeta]

I'm really disappointed by AMD, guess i'll have to go intel for the first time.

 

[Biggles 266]

I'm really disappointed by AMD, guess i'll have to go intel for the first time.

Why disappointed?!

You do realise that the Core architecture is not currently available, don't you? If you needed to build a system today, Socket 939 and Athlon 64 X2s are still the best you can buy. Yeah, the early Conroe benchmarks look good, but take them with a pinch of salt, nobody has had full production version access yet. We know nothing about their true overclocking potential or their real cost. We also don't know for sure what AMD will be doing when Conroe is fully released.

As we speak, AMD are still the performance kings. We can look forward to the future, but we cannot judge that which we do not know.

I feel this is more of marketing gimmick by AMD. There's really no use for such a technology except as a proof of concept. This type of technology will probably give developers an excuse to not producing fully multi-threaded and scalable programs - It's basically a step backwards. We should be developing programs which use threads so that programs will scale well. Also such a technology will probably not scale well as it would only be able to split a thread up so much. At the end of the day, there are very few cpu-intensive applications that are still single-threaded. Games are one of them however this is changing with multi-threaded drivers and games starting to appear. Music and Video compression algorithms are largely multi-threaded so are large applications like Photoshop. In conclusion - I wouldn't get too excited about it.

I don't think developers will use it as an excuse for poorly threaded code. If you consider professional level content creation programs, by that I mean something like Adobe Photoshop or 3Dmax, the sort of programs that people build big workstations for - then (aside from them already being multithreaded) developers are going to want to try and keep Intel users happy, and making programs that don't scale well with regards to extra cores would alienate Intel users.

The other kind of program that benefits from more CPU power is games. And for a games developer, it is basically shooting yourself in the foot to write your games in a way that is poorly threaded since it will make ports to the Xbox 360 and the PS3 considerably more difficult. Even for PC only developers, it would still alienate Intel users, which is never a good thing.

What this is good for is legacy applications. Older, or at least current, games which aren't multithreaded will benefit. Lots of old, but still usable, custom business software will benefit. Loads of stuff that currently depend upon a single thread will benefit. Even programs that are currently multithreaded will benefit; you can have a program that spawns 20 threads, but if 19 of them barely do anything and only one uses lots of CPU times, then you will still find that a multiple-core machine will not be used to its full potential. This will be because the operating system assigns a thread to a core; but if only one core does most of the work then the program will still benefit from split over two cores.

The thing is, it is incredibly difficult to write code that is really well threaded. Sure, you can write code that spawns extra threads here and there, and that is a good thing. But synchronisation issues are very much at the forefront of any coder who is doing concurrent programming. The other problem is how many cores too take advantage of? Two is a safe bet, but what about in the next year or two when four cores becomes more popular? Adding extra threads does create overheads - this anti-HT idea would give the benefits of adding extra threads but without the overheads that comes with it.

 

[Vegeta]

Mate conroe will woop amd ass its clear as daylight, I have an AMD64 already btw.