AMD's Stuff?

caeteez · 14 Jan 2007 at 12:55

Any one know when AMD's new stuff is coming out, and whether or not they might actually be faster than Intels new C2D products?

Jokester · 14 Jan 2007 at 13:04

AMD K8L based chips are supposed to be out later this year, my gut feeling is that they won't be faster than even current top end C2D based processor's despite offering quad cores that communicate over a HT link.

Given that Intel will likely have 45nm based processors coming out at the end of this year/early next year it's not something I'm getting excited about though benchmarks may chance that.

Jokester

caeteez · 14 Jan 2007 at 13:16

Thanks for the info.

I only ask as I have an AM2 mobo w/ Crossfire support, so would like to be upgrading to an R600 with whatever AMD bring out to counter Intel.

Purhaps I should suck up ~£150 more and get an Intel mobo?

jaykay · 14 Jan 2007 at 13:28

at this point in time i would till amd get the kl8 desgin out
keep the mobo as the kl8 design may run on your mobo with a bios update
buy the r600 when it comes out
if the kl8 design is poor then make the change to intel

Gashman · 14 Jan 2007 at 13:40

jaykay said:
at this point in time i would till amd get the kl8 desgin out
keep the mobo as the kl8 design may run on your mobo with a bios update
buy the r600 when it comes out
if the kl8 design is poor then make the change to intel

im with you on this one, for the simple fact that people think AMD aren't capable of besting conroe (not sure what brings them to that conclusion mind) but it leaves you with a way out if K8L doesn't perform as you want

locutus12 · 14 Jan 2007 at 14:42

jaykay said:
at this point in time i would till amd get the kl8 desgin out
keep the mobo as the kl8 design may run on your mobo with a bios update
buy the r600 when it comes out
if the kl8 design is poor then make the change to intel

K8L`s integer performance based on available reviews is already set to be far in excess of conroes and there are a few other things they may or may not do too but for the most part they are tight lipped. personally id say stick with what you have, unless your planning on encoding DVD`s day and night the only performance difference your going to see will only be able to be measured if your sat at your pc with a stop watch.

Jokester said:
AMD K8L based chips are supposed to be out later this year, my gut feeling is that they won't be faster than even current top end C2D based processor's despite offering quad cores that communicate over a HT link.

read here.

http://www.xbitlabs.com/articles/cpu/display/amd-k8l.html

Jokester · 14 Jan 2007 at 15:02

locutus12 said:
read here.

http://www.xbitlabs.com/articles/cpu/display/amd-k8l.html

Tells me nothing about real world performance.

Jokester

locutus12 · 14 Jan 2007 at 15:04

Jokester said:
Tells me nothing about real world performance.

Jokester

indeed, so why say this?

Jokester said:
AMD K8L based chips are supposed to be out later this year, my gut feeling is that they won't be faster than even current top end C2D based processor's despite offering quad cores that communicate over a HT link.

my information is based upon the actual design of the chip as opposed to a gut feeling.

Jokester · 14 Jan 2007 at 15:08

locutus12 said:
indeed, so why say this?

my information is based upon the actual design of the chip as opposed to a gut feeling.

My gut feeling is based on years of experience that bigger numbers doesn't necessarily mean faster hardware. As I said originally, until I see real world results then at the moment it's not something I'm going to be getting excited about or tell people to put off buying a new system now, on the off chance that K8L is vastly superior to Core2Duo, or more importantly what processors Intel have out at the time.

Jokester

caeteez · 14 Jan 2007 at 17:54

Well thanks for ya advice lads, dont get in a tiff over it though

Dolph · 14 Jan 2007 at 18:05

locutus12 said:
indeed, so why say this?

my information is based upon the actual design of the chip as opposed to a gut feeling.

You have a processing diagram or internal architecture diagram to share with us then? Because I've not been able to find one for the chip yet.

The only stuff I can find is stuff like this

http://www.digit-life.com/articles2/editorial/amd-guiseppe-amato-conf-part1.html

Which doesn't really tell us anything at all.

For an example of the sort of thing I'm looking for, look here

http://arstechnica.com/articles/paedia/cpu/core.ars/2

At the core microarchitecture diagram. This is the sort of diagram we need to see to make the sort of judgements about expected performance.

Gashman · 14 Jan 2007 at 22:17

Jokester said:
Tells me nothing about real world performance.

Jokester

very true, as long as AMD aren't bulling then maybe, just maybe its concievable to think K8L might over-take conroe (nightmare for the recently formed conroe fanclub

) but then i see sooo much rubbish floating around this forum like 'omfg i did super-pi in 15 seconds!', what does super-pi tell you about real world performance, absolutely nothing, all we know are intels core architecture likes super-pi, beating there AMD counterparts by 100% in a lot of cases, but is the processor 100% faster than AMDs, not even close so the only real world tests you should listen to are FPS in games and speeds of other applications, super-pi can show us the potential of a processor at number crunching, but nothing more

as he said though, AMD looking to significantly improve interger performance in K8L as well as having much much more aggressive pre-fetching and ways to minimise those stalls that occur due to loosing data, plus native quad core (not two conroes together like intels) all connected via HT links, hyper-transport 3.0 with more bandwidth, hopefully greater scalability so we can get some nice clock speeds, i don't think they'll fail to be honest, theres a lot resting on the shoulders of K8L, they can't afford it not to beat conroe

Dolph · 14 Jan 2007 at 22:27

This is why I want to see some actual information, I'm sure the basics on Conroe were out by now timescale-wise. I want to see what AMD are planning for each core (because simply welding four A64 cores together isn't going to cut it, even with HT linking), so I'm hoping they have something else up their sleeve.

So far I don't think there's enough information to pass any sort of judgement, good or bad, on what kind of performance we can expect from these chips. Those counting them out and those already counting on them beating conroe/kentsfield/yorkfield are all simply speculating.

drak3 · 15 Jan 2007 at 00:24

The less the nanometers get, the faster the processors.

Simply because everything gets closer.

Intel is miles ahead I think since they already got 45nm masks and printed circuits. Either AMD is hiding something or they lost the game.

Rroff · 15 Jan 2007 at 01:36

1. The best performance desktop AMD CPUs FX-62, etc. are ~25% slower clock for clock in real world applications to a conroe... a 3.6gig conroe is easily 50% faster than most current AMD CPUs will ever reach even with overclocking...

2. The K8L is in trouble... if it beats the conroe by a small margin no one will care* - especially as it will likely be £400+ at launch... if it beats the conroe by a large margin, no one will care much unless its competitvely priced against the conroe... and I don't see that happening with a newly released CPU... and if it does beat the conroe by a large margin and is competitively priced no one is going to care* either because by then intel will have moved on to better CPUs and the conroe will look as quaint as the Pentium D...

*except the rabid AMD fan club

locutus12 · 15 Jan 2007 at 12:04

drak3 said:
The less the nanometers get, the faster the processors.

Simply because everything gets closer.

Intel is miles ahead I think since they already got 45nm masks and printed circuits. Either AMD is hiding something or they lost the game.

Intel has always been ahead with the shrinking process, Intel hit 90nm way before AMD yet AMD still managed to beat intel for many years. Your argument isnt too valid, at least not yet.

Rroff said:
The K8L is in trouble... if it beats the conroe by a small margin no one will care* - especially as it will likely be £400+ at launch... if it beats the conroe by a large margin, no one will care much unless its competitvely priced against the conroe... and I don't see that happening with a newly released CPU... and if it does beat the conroe by a large margin and is competitively priced no one is going to care* either because by then intel will have moved on to better CPUs and the conroe will look as quaint as the Pentium D...

*except the rabid AMD fan club

and switch that little pile of bs around and you have the very same arguments people put against conroe. And they were quite wrong.

drak3 · 15 Jan 2007 at 17:22

Assume you got a conroe at 45nm to push to the limit? Intel's mistake was going for the clocks only so far. Since they realised that was wrong and designed a good chip from all sides, they made conroe.

ps: For us consumers it doesnt matter really tho

Just keep in mind that we buy what we can afford ( the majority) and thats the mid range chip which happens to be conroe for now as it was athlon before. As long a company holds the top in this market, its a winner. Thats my point of view anyway

Rroff · 15 Jan 2007 at 17:50

locutus12 said:
and switch that little pile of bs around and you have the very same arguments people put against conroe. And they were quite wrong.

whats bs about it? its hard logic, if you can't see that then I expect you'll be the first one spending £500+ on a CPU that is likely to only barely beat an E6600 in real world applications...

only rabid AMD fanatics put this argument against the conroe... the same ones who believe the K8L is gonna save the world... I can understand their stance when you look at the P4 and Pentium D... unfortunatly most of them didn't take a look at the Pentium-M (or even knew it existed) anyone who had used a Pentium-M and knew where the conroe ideas were coming from wouldn't have said that about the conroe...

Kamakazie! · 15 Jan 2007 at 22:10

Rroff said:
whats bs about it? its hard logic, if you can't see that then I expect you'll be the first one spending £500+ on a CPU that is likely to only barely beat an E6600 in real world applications...

only rabid AMD fanatics put this argument against the conroe... the same ones who believe the K8L is gonna save the world... I can understand their stance when you look at the P4 and Pentium D... unfortunatly most of them didn't take a look at the Pentium-M (or even knew it existed) anyone who had used a Pentium-M and knew where the conroe ideas were coming from wouldn't have said that about the conroe...

You do talk some rubbish.
"a CPU that is likely to only barely beat an E6600 in real world applications...". Based on who's judgement?
An A64 if much less than 25% slower than a conroe, clock for clock, across the board. More like 10-15% from all the reviews i have read.
People will buy whatever is the best processor for them, regardless of if one is slightly faster clock for clock. It comes down to many factors. Saying the K8L needs to be vastly faster than Conroe and competitive on price to sell is rubbish.

Dolph, if you want information (which you can't seem to find... though it is not hard) on K8L architecture take a look at the X-bit labs article already posted and this link. We have a die shot and all the technical data you need to make an informed decision. There is no guaranteeing that it will turn out to be correct but it is most likely.

DailyTech - Search for more articles

Plus others all over the web.

Rroff · 15 Jan 2007 at 22:52

Kamakazie! said:
More like 10-15% from all the reviews i have read.

If you have a very high end opteron or FX-62 (which incidentally cost well over twice the price of an E6600) then it may be closer to 15% with most of the consumer FX-62/60 models and the whole X2 range (which admittedly have come down a lot in price recently) its a good 25%.

Kamakazie! said:
Saying the K8L needs to be vastly faster than Conroe and competitive on price to sell is rubbish.

must be more fools in this world than I imagined... in even a few months, let alone 6+ months time no one in their right mind is going to buy a CPU that is only moderatly faster* than a current conroe and a lot more expensive...

* quad core will give them a sizeable lead over a core 2 duo in some areas

EDIT:

K8L Q3 2007 quad core with a clock speed of ~2.8gig... as I said conroe will look quaint by then, but that will be up against a 45nm Intel Core 2 Quad with speeds around ~3.7 gig...