Will AMD ever regain the performance crown again?

[FelixDeCat]

Originally posted by: DaveSimmons
Some years ATI wins, some years nvidia does, some years AMD wins, some years intel does. Some times you feel like a nut, some times you don't.

Almond Joys got nuts....Mounds dont. Will Ford beat Chevy again? Will Coke beat Pepsi?

Who cares.

 

[Viditor]

Furen:

The reason why K8 has more room for improvement is simply because it is crappier

I must admit, I like your answer better than mine...it's cleaner and kinda sums it all up.


Disclaimer: The opinions expressed by Viditor are his alone and do not necessarily represent the AT Forums or any of it's posters...

 

[dmens]

Originally posted by: Furen
Now, let me answer the question instead of letting Viditor answer for me:

The reason why K8 has more room for improvement is simply because it is crappier. The microarchitecture is pretty aged, if you think about it, as it is based on the K7. Sure, there were some improvements when the K8 was launched but these improvements did not really touch the backend at all and were relatively minor when taken as a whole. Kind of like how Yonah/Dothan/Banias were better performers than the P6 but nowhere near what Core accomplished. The memory controller IS big but AMD's buffers, data paths, L2 cache bandwidth, prefetchers, decode performance, execution unit performance, lack of load reordering and the like put the microarchitecture at a significant disadvantage. Intel's Core, on the other hand, does not have any glaring weaknesses you can point to. I'm sure there's still stuff that can be done to it to improve performance but you could say that it is the pinnacle of Intel tech RIGHT NOW.

EDIT: Regs, AMD has had micro-op fusion since the original K7 came out. If you mean MACRO-op fusion then I'm not sure how much of a benefit it really is. Just look at Core 2's x86-64 benchmarks, we don't see a huge weakness even though it lacks macro-op fusion in 64-bit mode.

there's no such thingh as a pinnacle of development on a core with the hybrid design style of modern cpu's. there is an arbitrary amount of performance that can be extracted. the only thing holding back is power and die size.

k8 isn't crappy, amd just hasn't taken big chances to mitigate risk and schedule. i dont think merom took too many risks either. there's a lot more than can be done on that design.

 

[Henny]

Intel has a built in negative feedback loop when they're in the lead.

They tend to get complacent, arrogant, lazy, and when this happens they get passed by without them even knowing it. They'll probably get that way over the next 1-2 years and when they do AMD will once again take the lead.

 

[MDme]

Originally posted by: Henny
Intel has a built in negative feedback loop when they're in the lead.

They tend to get complacent, arrogant, lazy, and when this happens they get passed by without them even knowing it. They'll probably get that way over the next 1-2 years and when they do AMD will once again take the lead.

did AMD do that too?

 

[Viditor]

Originally posted by: Henny
Intel has a built in negative feedback loop when they're in the lead.

They tend to get complacent, arrogant, lazy, and when this happens they get passed by without them even knowing it. They'll probably get that way over the next 1-2 years and when they do AMD will once again take the lead.

The problem I have with that analysis is that chip design is closer to a 5 year cycle, not a 1-2 year cycle. What I mean by that is that chips released over the next 1-2 years are already well past the halfway mark of development, so any "complacency" would have had to occured 2-3 years ago.

 

[Rottie]

AMD will get it back to the top no doubt about it

 

[StevenNevets]

I figure sense we're on the topic...

The next generation of AMD proccessors should work with an current AM2 motherboard, correct?

I plan on getting an M2N-E so I'd like to be positive

 

[Pabster]

Originally posted by: hennethannun
I'm not sure that C2d is quite 20% faster per clock than X2. the E6800 definitely outperforms the FX-62, but it is also clocked more than 100Mhz higher. I would say the performance advantage is probably 15-20% (so maybe 20% is the upper boundery, rather than the average). But you also have to remember that C2D is 65nm, while X2 is AMD is still on 90nm. Once K8 goes to 65, things will be slightly better (although i think C2d is still better than a 65nm X2).

Considering a stock 6600 beats an FX-62...

as for prices, where have you been? the 4400+ retails for just of $200 now. and they are widely available. AMD is still fine in the low to mid-end market (although things will be rough for them once the C2D market settles down). they have almost completely ceded the high-end and enthusiast market to Intel for the moment. Aside from the FX series, there is not an AMD chip on the market that retails for more than $400.

I believe he was referring to X2 launch prices and shortly thereafter. I paid upwards of $625 for an X2 4400+ right after launch. So a $350 mid-range Intel chip that destroys it is pretty incredible.

 

[zsdersw]

AMD chips don't really need any more memory bandwidth. They have plenty of it that's not being used. Any significant improvements in performance from AMD chips will be rooted in the architecture.

Considering that the Core 2 Duo is quite a bit faster than the K8 while having significantly less memory bandwidth.. I wonder what it would be like if it had the memory bandwidth afforded by HT. I suppose we won't see that until CSI. If CSI doesn't pan out for some reason, Intel can just sign on to HT; then AMD's biggest advantage becomes irrelevant.

 

[Viditor]

Originally posted by: zsdersw
AMD chips don't really need any more memory bandwidth. They have plenty of it that's not being used. Any significant improvements in performance from AMD chips will be rooted in the architecture.

Considering that the Core 2 Duo is quite a bit faster than the K8 while having significantly less memory bandwidth.. I wonder what it would be like if it had the memory bandwidth afforded by HT. I suppose we won't see that until CSI. If CSI doesn't pan out for some reason, Intel can just sign on to HT; then AMD's biggest advantage becomes irrelevant.

I tend to agree with you Z...which is why AMD's release of cHT protocol to the public domain came as quite a shock. However, a poster on another board pointed something out that made quite a bit of sense to me...even if Intel adopted HT, they would always be behind AMD's release schedule for improvements. Remember that using HT in a chip design isn't like tacking on a new device, it has to be integrated into the design itself. That would mean that Intel would have to use current HT designs and not future ones (unless they began to develop HT themselves), and it takes ~5 years to go from design to product.

 

[Pabster]

Except Intel doesn't need to adopt HT.

If they can continue to overcome the limits of a traditional FSB as they have with Core 2, the point is moot.

 

[tallman45]

Do you think Pepesi minds being #2 to Coke ? There is no #3

Same with AMD, as long as there is no #3, being #2 is pefectly fine

 

[harpoon84]

Originally posted by: zsdersw
Considering that the Core 2 Duo is quite a bit faster than the K8 while having significantly less memory bandwidth.. I wonder what it would be like if it had the memory bandwidth afforded by HT. I suppose we won't see that until CSI. If CSI doesn't pan out for some reason, Intel can just sign on to HT; then AMD's biggest advantage becomes irrelevant.

I don't think C2D's are that bandwith limited to be honest, as Anandtechs own review points out: http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2795&p=6

A 2.4% gain for a 25% FSB boost isn't a lot, and the law of diminishing returns suggests even if C2D had a 2GHz FSB you'll probably only end up with ~5% better performance overall.

 

[Viditor]

Originally posted by: Pabster
Except Intel doesn't need to adopt HT.

If they can continue to overcome the limits of a traditional FSB as they have with Core 2, the point is moot.

If Intel was to only produce 1-2P systems, I might agree with you. But there must be a good reason that Intel isn't releasing the C2D MP (4P+) chips for a year, and I suspect that part of that reason is platform bandwidth due to their current FSB model...

 

[zsdersw]

Originally posted by: Viditor
I tend to agree with you Z...which is why AMD's release of cHT protocol to the public domain came as quite a shock. However, a poster on another board pointed something out that made quite a bit of sense to me...even if Intel adopted HT, they would always be behind AMD's release schedule for improvements. Remember that using HT in a chip design isn't like tacking on a new device, it has to be integrated into the design itself. That would mean that Intel would have to use current HT designs and not future ones (unless they began to develop HT themselves), and it takes ~5 years to go from design to product.

They wouldn't have to be behind AMD's schedule at all. Do you think Intel hasn't been studying HT's current and future plans? Of course they have. They probably know as much about where HT is and where it's going in the future as anyone does and can design future chips to be ready for any new developments.. so that it can be as close to "tacked on" as possible.

 

[zsdersw]

Originally posted by: harpoon84
I don't think C2D's are that bandwith limited to be honest, as Anandtechs own review points out: http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2795&p=6

A 2.4% gain for a 25% FSB boost isn't a lot, and the law of diminishing returns suggests even if C2D had a 2GHz FSB you'll probably only end up with ~5% better performance overall.

It probably isn't bandwidth limited.. but that wasn't my point. The performance picture of Core 2 Duo with HyperTransport would probably look quite a bit different than it does with a bumping of the FSB.

 

[xtknight]

Nope...never. AMD gave up! In a year something from AMD will kill Conroe. Do you have that little faith in the CPU companies that one will never be able to beat the other ever again? You probably didn't think Intel would ever beat AMD. Alas, they did. It just takes time.

 

[hennethannun]

Originally posted by: Pabster

Originally posted by: hennethannun
I'm not sure that C2d is quite 20% faster per clock than X2. the E6800 definitely outperforms the FX-62, but it is also clocked more than 100Mhz higher. I would say the performance advantage is probably 15-20% (so maybe 20% is the upper boundery, rather than the average). But you also have to remember that C2D is 65nm, while X2 is AMD is still on 90nm. Once K8 goes to 65, things will be slightly better (although i think C2d is still better than a 65nm X2).

Considering a stock 6600 beats an FX-62...

as for prices, where have you been? the 4400+ retails for just of $200 now. and they are widely available. AMD is still fine in the low to mid-end market (although things will be rough for them once the C2D market settles down). they have almost completely ceded the high-end and enthusiast market to Intel for the moment. Aside from the FX series, there is not an AMD chip on the market that retails for more than $400.

I believe he was referring to X2 launch prices and shortly thereafter. I paid upwards of $625 for an X2 4400+ right after launch. So a $350 mid-range Intel chip that destroys it is pretty incredible.

For the first part of your statement, yes, a stock 6600 does beat an FX-62 in most benchamarks, but not by all that much, they are pretty similar overall. And the clock different between the 20 is about 15% (2.4ghz is about 85% of 2.8ghz). so i think 15-20% better on clock per clock basis is about accurate.

as for the second part of your post, you are correct, he was talking about the initial prices for X2, but i answered that argument in a later post.
AMD was introducing dual-core to the home desktop environment. Core 2 is entering a market saturated with dual-core solutions. Also, Intel has a lot more production capacity than AMD, and can afford to offer ridiculous MSRPs for performance parts. And finally, if you look at the *REAL* prices for C2D parts, they are almost 20% higher than MSRP across the board (except for the E6600, which is even more inflated). I think the only real complaint about AMD's X2 pricing strategy was that from the introduction of the line last summer until just before C2D launched, the price of owning an X2 only dropped about $50. they did add the 3800+ to the bottom of the line, but each individual part was only about $50 cheaper in July of 2006 than it was inJuly of 2005.

 

[bob661]

No.

 

[wwswimming]

Originally posted by: FelixDeKat

Originally posted by: DaveSimmons
Some years ATI wins, some years nvidia does, some years AMD wins, some years intel does. Some times you feel like a nut, some times you don't.

Almond Joys got nuts....Mounds dont. Will Ford beat Chevy again? Will Coke beat Pepsi?

Who cares.

that almost sounds like a Cat Stevens song. or Merle Haggard. or Bonnie Raitt. or Ani DiFranco.

leapfrogs in performance has been the history of the industry.

i think the answer is going to be related to energy, the energy to power the devices & the energy to make them.

guess i'll just be using my old computers as heaters during the winter.

 

[Pabster]

Originally posted by: hennethannun
AMD was introducing dual-core to the home desktop environment. Core 2 is entering a market saturated with dual-core solutions. Also, Intel has a lot more production capacity than AMD, and can afford to offer ridiculous MSRPs for performance parts. And finally, if you look at the *REAL* prices for C2D parts, they are almost 20% higher than MSRP across the board (except for the E6600, which is even more inflated). I think the only real complaint about AMD's X2 pricing strategy was that from the introduction of the line last summer until just before C2D launched, the price of owning an X2 only dropped about $50. they did add the 3800+ to the bottom of the line, but each individual part was only about $50 cheaper in July of 2006 than it was inJuly of 2005.

Yep, and it would still be that way...if it weren't for the killer called Conroe Ain't competition grand?

 

[lenjack]

As others have commmented, it's cyclical. If you look back over the years, you'll see the performance crown shifting back and forth.

 

[Hard Ball]

Originally posted by: dmens

i interpret what you said as c2d is restricted by the lack of p2p links, which is bs. fyi, i refer to all c2d as merom. there is a 4p woodcrest on the market right now, mac pro.

What have you been smokin' ?? We all want some of what you have.

Mac Pro is 2P4C.
From the tech spec page of Mac Pro: http://www.apple.com/macpro/specs.html

Processing
Two 2GHz, 2.66GHz, or 3GHz Dual-Core Intel Xeon 5100 series processors
Intel Core microarchitecture
4MB shared L2 cache per processor
128-bit SSE3 vector engine
64-bit data paths and registers
Lower power optimization
1.33GHz, 64-bit dual independent frontside buses

didn't xbitlabs did some tests on memory throughput on c2d, and the conclusion is that it is not high impact at all, as expected. there are so many more annoying bottlenecks than the bus. not to mention the bus bottleneck has the most options available to work around, as opposed to many fundamental chokes in the core.

C2D is not a dual socket model, so whatever the results on C2D is completely irrelevant to this discussion; since there will be no coherency traffic outside of L2. Any effect of cache coherency will only affect L2 latency; and that's about it. You need to mention results of Xeon 51xx to mean anything for this discussion on FSB bottleneck.

link on useful plugins, please. is the cHT protocol even in use? if not, how would any plugin available now be any better than any other device using the pci bus?

cHT????

I think you mean ccHT. Really, what have you been smoking today??

 

[TheRyuu]

The performance crown between AMD and Intel is just like the government.

Right now the Republicans are Intel and have control of both the Senate and the Presidentry (SP?). AMD is the Democrats who are the minority right now. Overtime it goes back and forth, back and forth. AMD will surly come back eventually, it's just a matter of time.

Just because they lost this fight, doesn't mean they lost the never ending war.