AMD sent Techreport an Intel system to test their GPU's

[TerryMathews]

Not sure what you are trying to say. Because your original statement is still invalid. AMD took all the money they could when they had the lead. And in no way did they give any "charity". It was first in july 2006 we got a huge price change when AMD lost the performance lead. And we had 330$ E6600.

That's an interesting take. To me, it looks like Intel has SKUs that were both slower and more expensive.

I mean realistically why did AMD gain market share? How do you explain it beyond higher IPC at lower prices? We know the top end is a very small percentage of units sold. We also know that AMD wasn't dumping product. So doesn't that only leave superior price performance ratio?

 

[TerryMathews]

Not sure what you are trying to say. Because your original statement is still invalid. AMD took all the money they could when they had the lead. And in no way did they give any "charity". It was first in july 2006 we got a huge price change when AMD lost the performance lead we got cheapish CPUs again with the E6600. And that was simply because Intel understood more sold CPUs was better than higher prices in terms of profit. And AMD had to cut prices in the following months with above 75%.

Ah, there's the problem. You put words into my mouth.

I never said AMD was a charity. Or acted like one.

 

[Keysplayr]

And? http://www.sharkyextreme.com/guides..._3/Weekly-CPU-Prices--Week-of-May-23-2005.htm

Two words: Emergency Edition.

Terry. Your argument is done. It's obviously been that shown AMD was at least as expensive when they were on top.
And AMD gained market share at that time because they had the better product. It sold itself in reviews always besting equivilent P4's in everything but media encoding or some such.
So, I got no idea what you're trying to broadcast to us here either. ::shrugs::
And the "I never said" thing only goes so far. You can imply an infinite amount of things without actually saying it.

 

[TerryMathews]

Terry. Your argument is done. We've shown AMD was at least as expensive when they were on top.

Funny. I see a $900 and $1000 sku on that sheet. Don't you?

 

[mrmt]

They're both guilty of it at various points. Remember Intel selling the same core twice as the Pentium 2 and iii? I mean it was even the same speeds.

It was you who implied that AMD was *always cheaper* when clearly that isn't the case:

AMD was cheaper even when they held the performance crown.

AMD wasn't always cheaper than Intel and they did milk customers when they could. Pointing out that Intel did/does the same won't make your affirmation more correct.

 

[Rvenger]

AMD still made low cost solutions that with overclocking they could exceed the FX-55 speeds. Remember Opteron 144 and 146?

 

[Keysplayr]

Funny. I see a $900 and $1000 sku on that sheet. Don't you?

Like FX-55? Or even the X2 4800 when it was the top dog? Sure do.

 

[Keysplayr]

AMD still made low cost solutions that with overclocking they could exceed the FX-55 speeds. Remember Opteron 144 and 146?

So did Intel.

 

[ShintaiDK]

AMD still made low cost solutions that with overclocking they could exceed the FX-55 speeds. Remember Opteron 144 and 146?

You could OC Core 2 as well. E6400 anyone?

50% faster than FX62 and just as fast as X6800 for 1/4th the price. And still cheaper than an X2 4400+. And faster at stock too.

 

[Hitman928]

Ah, I see now. I had not connected the dots like that, despite knowing of all those events on an individualized basis.

Yes, what you say makes sense. In this light there is a pattern to the "disavow, discredit, deny" brigade.

What I guess I don't get is why people wouldn't want to see the products improved upon if such opportunities for improvement really do exist.

Take the more recent "high speed camera" video work published by TR. To be honest I saw hitching and stuttering from both video cards, more-so from the AMD card though, which I would hope Nvidia and AMD figure out a way to eliminate in future driver or hardware revisions.

If the customer willingly ignores the product deficiency, as they did in the early days of SSD adoption, then what company is going to intentionally spend money to shore up such deficiencies in future products?

It reminds me of the 24fps issues with Intel iGPUs when used as HTPCs...the "feature" is there, and it will always be there until Intel feels it necessary to spend the money needed to have its engineers fully, and finally, address the deficiency. Who doesn't want that happen? How is it going to happen if review sites do not put a spotlight on the existence of the issue in the first place?

These sorts of "consumer reports" ought to find unanimous common ground across enthusiasts in all segments IMO. It is when they don't that I become perplexed. As you rightly point out this is not an isolated incident when it comes to AMD products in general, for some reason AMD products are supposed to be granted immunity from scrutiny I guess.

Sad, really, because we all stand to benefit from an AMD that fields improved products, and who better to tell them what needs improving than the very people who are saying "I won't buy your product because of deficiency xyz"?

Not to spill over from VC&G, but most people posting about this on this forum were'nt doing so to defend AMD, but rather wanting to have techreport's methods validated. After all, the situation was exactly flipped last generation where the Amd cards were slower, but had much "smoother" frametimes, yet no one really seemed to care.

Additionally, there were some inconsistancies in their findings that people wanted investigated as well as how accurate fraps is at measuring this stuff anyway. Especially when Ryan from anandtech, as well as a few other very respected reviewers basically said that fraps is not a reliable way of doing what techreport is doing. The high speed camera was great, but when done on only one game whose engine is known to stutter (like FC3), it is not quite enough to prove a new methodology.

It's like if the DV team tells design that they have a new method to detect parasitic transistors even before fab, despite other experts saying the tools aren't reliable, you'd want them to thoroughly prove their methodology before relying on them. I for one welcome what techreport is doing, but want a bit more confirmation of their accuracy and relevance before really trusting their results.

BTW, great work with the 8350. The one thing I like about AMD chips compared to intel right now (because intel's obviously win on almost every metric), is the need to tweak them. To some, this might be a pain, but to an old(er) school overclocker who remembers all the tricks and tweaks that we used to have to do to overclock, I personally like the challenge, as compared to the simplicity of overclocking intel chips right now (excluding extreme cooling obviously). If I had the money to burn, I would have loved to pick up a piledriver cpu, just to play with.

 

[Hitman928]

You could OC Core 2 as well. E6400 anyone?

50% faster than FX62 and just as fast as X6800 for 1/4th the price.

Well, if we're talking core2, then yes, but FX55 was a generation before that when AMD ruled the scene, even if it was fairly short lived.

 

[ShintaiDK]

Well, if we're talking core2, then yes, but FX55 was a generation before that when AMD ruled the scene, even if it was fairly short lived.

I think his point was that you didnt have to buy the fastest AMD when AMD had the performance crown. Else his tatement didnt make much sense. Nether did you with Intel.

 

[mrmt]

Not to spill over from VC&G, but most people posting about this on this forum were'nt doing so to defend AMD, but rather wanting to have techreport's methods validated. After all, the situation was exactly flipped last generation where the Amd cards were slower, but had much "smoother" frametimes, yet no one really seemed to care.

If nobody complained in the last generation when the situation was flipped, is this an excuse to not bring up the issue now regardless of who is better at this metric? What should we do? Wait for a Nvidia-stuttering generation to raise the issue again?

First thing we have to determine the degree of relevance of this stuttering, and second a good methodology to measure it. Sure, we aren't already there but what TR is doing, raising the issue, is the right thing to do. The community will discuss and experiment until they can reach a standard measurement procedure.

 

[Hitman928]

If nobody complained in the last generation when the situation was flipped, is this an excuse to not bring up the issue now regardless of who is better at this metric? What should we do? Wait for a Nvidia-stuttering generation to raise the issue again?

First thing we have to determine the degree of relevance of this stuttering, and second a good methodology to measure it. Sure, we aren't already there but what TR is doing, raising the issue, is the right thing to do. The community will discuss and experiment until they can reach a standard measurement procedure.

I already said I think what techreport is doing is good and that I support their efforts, I only mentioned last gen because if no one cared then, it leads me to wonder if it's really an issue or if it's a solution in search of a problem. That's why I'm waiting for more investigations from TR and others before making any sort of conclusion or even real theories and look forward to the conclusion.

 

[TerryMathews]

If we're going to have a serious discussion about this, let's look at facts.

The best IPC estimate I can find on these processors is 2.25 for P4 and 3.0 for A64. Anyone have better numbers?

Analyzing the Intel OEM SKUs against AMD's price list, I find that AMD offers equivalent price/performance ratio to vastly superior (30.5%).

My methodology is IPC * clock speed / price from Sharky sheet to yield instructions per dollar.

That 30.5% is a 2800+ vs an OEM 3.0E

 

[TerryMathews]

Like FX-55? Or even the X2 4800 when it was the top dog? Sure do.

Logically, why did AMD gain market share during the K7 and early K8 period?

 

[Hitman928]

I think his point was that you didnt have to buy the fastest AMD when AMD had the performance crown. Else his tatement didnt make much sense. Nether did you with Intel.

eh? Unless I'm mistaken, the original idea was that when AMD had the performance lead, they were still cheaper than intel, someone pointed out the near $1000 cpu's (FX55) and his comment was that you could get a cheaper processor at the time for and overclock it to beyond FX55 speeds (Then you would have a cpu that performed better than intel's best at a cheaper price than intel, at the time). You responded with, yeah but what about core2 compared to FX62. But at that time, intel had re-taken the performance crown and AMD's freefall had begun, so I didn't see the relevance of that post in context to his and the timeframes didn't match up. Unless I missed someone's point. . .

 

[ShintaiDK]

eh? Unless I'm mistaken, the original idea was that when AMD had the performance lead, they were still cheaper than intel, someone pointed out the near $1000 cpu's (FX55) and his comment was that you could get a cheaper processor at the time for and overclock it to beyond FX55 speeds (Then you would have a cpu that performed better than intel's best at a cheaper price than intel, at the time). You responded with, yeah but what about core2 compared to FX62. But at that time, intel had re-taken the performance crown and AMD's freefall had begun, so I didn't see the relevance of that post in context to his and the timeframes didn't match up. Unless I missed someone's point. . .

Is the FX55 CPU all we use to determine if one is cheaper or not?

 

[Hitman928]

If we're going to have a serious discussion about this, let's look at facts.

The best IPC estimate I can find on these processors is 2.25 for P4 and 3.0 for A64. Anyone have better numbers?

Analyzing the Intel OEM SKUs against AMD's price list, I find that AMD offers equivalent price/performance ratio to vastly superior (30.5%).

My methodology is IPC * clock speed / price from Sharky sheet to yield instructions per dollar.

That 30.5% is a 2800+ vs an OEM 3.0E

IPC is highly workload dependent and usually completely misunderstood. I don't know why this even really matters, it's a decade old discussion and not really relevant to thread, I say we just drop it and move on

 

[TerryMathews]

eh? Unless I'm mistaken, the original idea was that when AMD had the performance lead, they were still cheaper than intel, someone pointed out the near $1000 cpu's (FX55) and his comment was that you could get a cheaper processor at the time for and overclock it to beyond FX55 speeds (Then you would have a cpu that performed better than intel's best at a cheaper price than intel, at the time). You responded with, yeah but what about core2 compared to FX62. But at that time, intel had re-taken the performance crown and AMD's freefall had begun, so I didn't see the relevance of that post in context to his and the timeframes didn't match up. Unless I missed someone's point. . .

My point specifically was that during the time that AMD had a performance lead over Intel, the K7 and early K8 days, AMD was cheaper than Intel. And when I say cheaper, I specifically mean a better price to performance ratio. Without over clocking on either side.

I'm not saying AMD was a charity, or that they didn't charge a premium for their premium products just like Intel. They did however offer a better value. The numbers don't lie. And logically if you think it makes sense. Why else would they gain market share? Certainly not brand recognition or superior marketing.

 

[Hitman928]

Is the FX55 CPU all we use to determine if one is cheaper or not?

I really don't understand the point you're trying to make.

 

[Lonyo]

If we're going to have a serious discussion about this, let's look at facts.

The best IPC estimate I can find on these processors is 2.25 for P4 and 3.0 for A64. Anyone have better numbers?

Analyzing the Intel OEM SKUs against AMD's price list, I find that AMD offers equivalent price/performance ratio to vastly superior (30.5%).

My methodology is IPC * clock speed / price from Sharky sheet to yield instructions per dollar.

That 30.5% is a 2800+ vs an OEM 3.0E

A better question would be:
WHY are we having a discussion about this.

 

[Rvenger]

So did Intel.

Netburst was garbage and this is irrelevant to this thread. Lets get back on topic.

 

[mrmt]

And when I say cheaper, I specifically mean a better price to performance ratio. Without over clocking on either side.

Thank god you clarified this point, because last time I checked a dictionary cheap meant "low price", not "good performance ratio".

Regardless of *YOUR* opinion using *YOUR* arbitrary numbers, there were times in which AMD chips were more expensive than Intel chips. There is nothing wrong with that, companies exist for the sake of make money and if they have a better product and don't sell for the best price they can get the executive team should be ousted ASAP. It is fair to say that AMD did what they had to do with the A64, and they will do so again if they can.

Just look at what they did with their 7000 series. Instead of keep same price levels and give NVDA a run for its money, they priced 7000 series so high that they didn't even gain market share. Was this a wrong strategy? Looking at AMD current numbers we can say it was. Was this unethical? Not at all. Nobody is obliged to buy AMD GPUs, you pay for what you want.

 

[TerryMathews]

Thank god you clarified this point, because last time I checked a dictionary cheap meant "low price", not "good performance ratio".

Regardless of *YOUR* opinion using *YOUR* arbitrary numbers, there were times in which AMD chips were more expensive than Intel chips. There is nothing wrong with that, companies exist for the sake of make money and if they have a better product and don't sell for the best price they can get the executive team should be ousted ASAP. It is fair to say that AMD did what they had to do with the A64, and they will do so again if they can.

Just look at what they did with their 7000 series. Instead of keep same price levels and give NVDA a run for its money, they priced 7000 series so high that they didn't even gain market share. Was this a wrong strategy? Looking at AMD current numbers we can say it was. Was this unethical? Not at all. Nobody is obliged to buy AMD GPUs, you pay for what you want.

Orly? Because we both looked at the same sheet that had a $1000 Intel chip on it.

Lol @ you calling price performance ratio arbitrary. Fanboi.