Does anyone have one of these rarities? The story goes that AMD was set to release the K6 by the time the K5 was ready to hit 200mhz so only small numbers of these got out so they wouldn't draw sales away from the K6-200.
I believe your assumption would be correct. Afaik it was only released in some country in east Asia.
I'm not sure I even have a any K5's kicking around somewhere here though.
I have a K5 PR200. I bought it off of CPUWorld when they did a bulk order years ago.
I tested this CPU in my 133 MHz Challenge thread. PR200 runs at 133 MHz.
wrote:I have a K5 PR200. I bought it off of CPUWorld when they did a bulk order years ago.
I tested this CPU in my 133 MHz Challenge thread. PR200 runs at 133 MHz.
Why does it look like per clock performance of the K5 FPU is superior to the K6 FPU? If the K5 FPU was faster, why didn't AMD use it in the K6?
They are very different CPUs. As far as I know they have nothing in common besides using the same socket and being x86. K5 is a beast at per clock performance but it can't go much beyond 133 MHz. It was a dead end.
wrote:They are very different CPUs. As far as I know they have nothing in common besides using the same socket and being x86. K5 is a beast at per clock performance but it can't go much beyond 133 MHz. It was a dead end.
But what I'm saying is why didn't they use just the FPU design from the K5? If they had done that then maybe the whole K6 family would have had a lot better performance in games and been able to keep up with Intel.
wrote:wrote:They are very different CPUs. As far as I know they have nothing in common besides using the same socket and being x86. K5 is a beast at per clock performance but it can't go much beyond 133 MHz. It was a dead end.
But what I'm saying is why didn't they use just the FPU design from the K5? If they had done that then maybe the whole K6 family would have had a lot better performance in games and been able to keep up with Intel.
You can't just plug some FPU into another chip and expect it to work. K6 was based on Nextgen iirc while K5 was a totally different chip. K5 and K6 are totally unrelated.
And besides, the reason why AMD abandoned K5 is that, even though it may run fast per clock tick, this is meaningless if you can't get your chip to scale. And that's exactly what happened to K5
wrote:I have a K5 PR200. I bought it off of CPUWorld when they did a bulk order years ago.
I tested this CPU in my 133 MHz Challenge thread. PR200 runs at 133 MHz.
Yup, gotta love those group buys! 😁
wrote:You can't just plug some FPU into another chip and expect it to work. K6 was based on Nextgen iirc while K5 was a totally differ […]
wrote:wrote:They are very different CPUs. As far as I know they have nothing in common besides using the same socket and being x86. K5 is a beast at per clock performance but it can't go much beyond 133 MHz. It was a dead end.
But what I'm saying is why didn't they use just the FPU design from the K5? If they had done that then maybe the whole K6 family would have had a lot better performance in games and been able to keep up with Intel.
You can't just plug some FPU into another chip and expect it to work. K6 was based on Nextgen iirc while K5 was a totally different chip. K5 and K6 are totally unrelated.
And besides, the reason why AMD abandoned K5 is that, even though it may run fast per clock tick, this is meaningless if you can't get your chip to scale. And that's exactly what happened to K5
wrote:I have a K5 PR200. I bought it off of CPUWorld when they did a bulk order years ago.
I tested this CPU in my 133 MHz Challenge thread. PR200 runs at 133 MHz.
Yup, gotta love those group buys! 😁
The design HAD to be better than the one that Nexgen was using because Nexgen was still using the 387 as the basis of it's FPU. If Nexgen could get a 387 to scale then AMD should have had no problem getting their own FPU design to scale.
Well the K5 is some sort of hybrid of the AMD 29000 RISC processor. Basically it translates x86 into its internal language. This is like Pentium Pro and K6 but they are all different. K6 was Nexgen's RISCy CISC 😁
K6 was probably used basically as it was acquired. It takes years to redesign CPUs.
Besides if you want to see x87 FPU done right then all you need to do is divert your attention to Athlon.
Also, K6's game performance isn't entirely the fault of its FPU. It also has terrible memory performance compared to any P6 chip. This is also part of why Athlon didn't totally dust P3. Athlon at least had a separate cache bus but of course it was still saddled with horrid chipsets until nForce 2.
wrote:The design HAD to be better than the one that Nexgen was using because Nexgen was still using the 387 as the basis of it's FPU. If Nexgen could get a 387 to scale then AMD should have had no problem getting their own FPU design to scale.
Yes, but Nexgen was using the nx586 and AMD's K6 is based on the nx686. In other words, no, it was definitely not better.
If it was as you say they were, then why did a multimillion dollar company like AMD base it's K6 on a chip design that's totally unrelated to K5?
I think you'd better do a bit of reading as your facts are way off.
Here is what happened to Nexgen, and here is how K6 was born.
sliderider, don't forget that AMD ditched K5 because it was the design of K5 that prevented AMD from keeping up with Intel's chips. Like swaaye said, it was a dead end.
You can create a super duper good chip that performs 5 times better then i7, but that won't do you any good if that CPU can't be clocked past 300Mhz, it's really that simple.
Or... you can create a real shitty one like Netburst, and hope it'll sclae to 10ghz!!!
Actually if you think about it P4 was more competitive against A64 than Phenom II is right now against i3-i7. 😉 And also modern quad / hexa cores are about as hot as Prescott but we have much better heatsinks now and the Cool'n'Quiet/Speedstep stuff really reigns in the off peak power.
Edit: Woops...wrong topic!!! 😁
Edit2:9am....it shows 😜
Anyway, I'd love to get my hands on a PR200 or at least a PR166. I actually ordered 2 such chips from cpu-world but apparently the seller decided to quit the scene, leaving people who actually paid for their chips with empty hands.
Good thing I never received my invoice or I might've been left in limbo as well.
It's not a top priority for me though, perhaps one will eventually fall into my lap.
wrote:Or... you can create a real shitty one like Netburst, and hope it'll sclae to 10ghz!!!
LMAO!!
wrote:Actually if you think about it P4 was more competitive against A64 than Phenom II is right now against i3-i7. 😉 And also modern quad / hexa cores are about as hot as Prescott but we have much better heatsinks now and the Cool'n'Quiet/Speedstep stuff really reigns in the off peak power.
I'm not so sure though. After a quick search I found the i7 has a TDP of around 130W while most Phenom II's have a TDP of 125W.
Back in the Netburst/A64 days the P4's TDP was higher then A64's one while P4 was still slower.
Things were so bad, Intel decided to not even bother with Tejas anymore but otoh AMD did release Thuban.
wrote:Yes, but Nexgen was using the nx586 and AMD's K6 is based on the nx686. In other words, no, it was definitely not better. […]
wrote:The design HAD to be better than the one that Nexgen was using because Nexgen was still using the 387 as the basis of it's FPU. If Nexgen could get a 387 to scale then AMD should have had no problem getting their own FPU design to scale.
Yes, but Nexgen was using the nx586 and AMD's K6 is based on the nx686. In other words, no, it was definitely not better.
If it was as you say they were, then why did a multimillion dollar company like AMD base it's K6 on a chip design that's totally unrelated to K5?
I think you'd better do a bit of reading as your facts are way off.
Here is what happened to Nexgen, and here is how K6 was born.sliderider, don't forget that AMD ditched K5 because it was the design of K5 that prevented AMD from keeping up with Intel's chips. Like swaaye said, it was a dead end.
You can create a super duper good chip that performs 5 times better then i7, but that won't do you any good if that CPU can't be clocked past 300Mhz, it's really that simple.
You're not comprehending what I'm saying. I know the history of NexGen and AMD so I don't need your links. The reason AMD bought Nexgen was because the K5 wouldn't scale and their own 686 design was going nowhere. I know that already. NexGens FPU design was clearly inferior to the one AMD already had, though. If you do an apples to apples comparison and put a AMD K5 PR120 against a Nexgen PF120, the FPU in the Nexgen loses. What makes it even worse for the Nexgen was the fact that the AMD K5 PR120 is actually clocked at 90mhz while the PF120 is clocked at about 108, so how can the AMD K5 FPU running faster at a slower clock speed be inferior to the Nexgen? AMD could have coupled the FPU in the K5 with the integer core of the Nexgen and it would have been a better chip than the K6 that we got and been more competitive with Intel. Do you understand now?
For proof of this, all you have to do is go to swaayes 133 mhz challenge and you can see for yourself that at the same clock speed the K5 is roughly 36% faster than the K6 at FPU calculations.
wrote:NexGens FPU design was clearly inferior to the one AMD already had, though. If you do an apples to apples comparison and put a AMD K5 PR120 against a Nexgen PF120, the FPU in the Nexgen loses. What makes it even worse for the Nexgen was the fact that the AMD K5 PR120 is actually clocked at 90mhz while the PF120 is clocked at about 108, so how can the AMD K5 FPU running faster at a slower clock speed be inferior to the Nexgen?
Dude, I only said K5 was inferior to K6. And bringing up nx586 is completely irrelevant as it has nothing to do with AMD.
wrote:AMD could have coupled the FPU in the K5 with the integer core of the Nexgen and it would have been a better chip than the K6 that we got and been more competitive with Intel. Do you understand now?
For proof of this, all you have to do is go to swaayes 133 mhz challenge and you can see for yourself that at the same clock speed the K5 is roughly 36% faster than the K6 at FPU calculations.
Dude, if you were to cut out the FPU of the K5 and plump it into a K6 somehow, your K6 would probably again be limited to K5's maximum speed, which was 133Mhz.
AMD didn't abandon K5 for no reason you know.
Like I mentioned before, if you have a CPU that's 50% faster for each clock cycle but runs at half the speed, you're still doing a step backwards.
If you know the background story of the nx686 and the K6, then wouldn't you already know yourself why AMD didn't do exactly what you've mentioned?
I'll tell ya. AMD was falling back in the Mhz race and had to get something on the shelves very quickly. If they had done what you say, it would've set AMD even further back then it already was. AMD needed to generate cash fast. That's why they used nx686 as a base for their K6.
Making CPU's is not about being the fastest or making the most efficient chip, it's about making money and designing a K5-FPU + nx686 CPU wasn't the money way.
wrote:Making CPU's is not about being the fastest or making the most efficient chip, it's about making money and designing a K5-FPU + nx686 CPU wasn't the money way.
Hit the nail on the head...
wrote:Dude, I only said K5 was inferior to K6. And bringing up nx586 is completely irrelevant as it has nothing to do with AMD. […]
wrote:NexGens FPU design was clearly inferior to the one AMD already had, though. If you do an apples to apples comparison and put a AMD K5 PR120 against a Nexgen PF120, the FPU in the Nexgen loses. What makes it even worse for the Nexgen was the fact that the AMD K5 PR120 is actually clocked at 90mhz while the PF120 is clocked at about 108, so how can the AMD K5 FPU running faster at a slower clock speed be inferior to the Nexgen?
Dude, I only said K5 was inferior to K6. And bringing up nx586 is completely irrelevant as it has nothing to do with AMD.
wrote:AMD could have coupled the FPU in the K5 with the integer core of the Nexgen and it would have been a better chip than the K6 that we got and been more competitive with Intel. Do you understand now?
For proof of this, all you have to do is go to swaayes 133 mhz challenge and you can see for yourself that at the same clock speed the K5 is roughly 36% faster than the K6 at FPU calculations.
Dude, if you were to cut out the FPU of the K5 and plump it into a K6 somehow, your K6 would probably again be limited to K5's maximum speed, which was 133Mhz.
AMD didn't abandon K5 for no reason you know.
Like I mentioned before, if you have a CPU that's 50% faster for each clock cycle but runs at half the speed, you're still doing a step backwards.If you know the background story of the nx686 and the K6, then wouldn't you already know yourself why AMD didn't do exactly what you've mentioned?
I'll tell ya. AMD was falling back in the Mhz race and had to get something on the shelves very quickly. If they had done what you say, it would've set AMD even further back then it already was. AMD needed to generate cash fast. That's why they used nx686 as a base for their K6.Making CPU's is not about being the fastest or making the most efficient chip, it's about making money and designing a K5-FPU + nx686 CPU wasn't the money way.
Not necessarily. How do you know that it was the FPU design that was holding it back? The FPU design could have been more scalar than the rest of the chip. The CPU and FPU are technically two different chips on the same silicon so the limitations that apply to one would not necessarily apply to the other.
wrote:wrote:Making CPU's is not about being the fastest or making the most efficient chip, it's about making money and designing a K5-FPU + nx686 CPU wasn't the money way.
Hit the nail on the head...
Oh, and what about all the sales they lost to Intel because their floating point performance sucked? Wouldn't there have been more money in grabbing those sales away from Intel by including a better FPU even if it took a little more time to incorporate it into the design?
wrote:wrote:wrote:Making CPU's is not about being the fastest or making the most efficient chip, it's about making money and designing a K5-FPU + nx686 CPU wasn't the money way.
Hit the nail on the head...
Oh, and what about all the sales they lost to Intel because their floating point performance sucked? Wouldn't there have been more money in grabbing those sales away from Intel by including a better FPU even if it took a little more time to incorporate it into the design?
And how long would you reckon this "little more time" you speak of would exactly take?
