Question on processor speed.....

[webe3]

OK, I have a AMD Athlon XP 2000 chip, but the ACTUAL clock speed is ONLY 1.67 ghz instead of 2 ghz. (Now I understand the theory of how a 12 speed can overtake a regular bicycle when it changes gears even though the regular bike is faster) But here's my question.....I am kinda puzzled at these new processors that AMD has come out with! Take their "flagship" processor The AMD Barton 3200. The 3200 is REALLY misleading, because instead of even "comming close" in core memory like my xp 2000....the 3200's core memory is ONLY 2.2 ghz!! But you are talking about a processor that is SUPPOSED to be able to OUTPERFORM a Intel Pentium 4 3.2 ghz chip?? That seems like it is a stretch! THAT IS A 1 GIGAHERTZ LACK OF CORE MEMORY BETWEEN THE TWO! Now I know that in my athlon xp 2000, they can make it perform better than a regular Pentium 4 2 ghz chip by increasing the chips overall performance speed.....but to market a chip and say it will perform as good or better than a Intel 3.2 Ghz and you CALL your chip the 3200....BUT are only working with 2.2 ghz of stock speed!?? That seems to be an impossible task! I can see how a xp 2000 can keep up with a Intel 2 ghz because there is only a 300 mhz difference between the two chips, BUT HOW CAN YOU JUSTIFY CALLING A AMD 3200 CHIP THE SAME AS THE INTEL 3.2 ghz????? It just seems too far of a stretch if you ask me! By the way, I know you can also overclock your processor as mine would be easy to overclock to 2ghz....but I have never HEARD of anyone being able to overclock a chip 1ghz worth!! Which is what you would need to do to bring it up to speed with the 3.2 Intel. It just doesn't make sense to me!

[zeradul]

The 3200+ rating is a direct mathematical computation where they take computations per cycle multiplied by megahertz.

So with your bike analogy, its like a bike with larger diameter tires don't need to spin as fast to go the same speed.

[minibubba]

ever heard of the 'megahertz myth'?

megahertz isn't everything :devil win

[webe3]

Guys that is all probably true, BUT YOU STILL HAVEN'T EXPLAINED ABOUT THE 1 GHZ CORE MEMORY DIFFERENCE! Until somebody can explain THAT one to me, I am going to have a hard time buying it!

[webe3]

ever heard of the 'megahertz myth'?

megahertz isn't everything :devil win

YES I have heard of that myth, BUT COME ON! A 1 ghz difference in core memory IS NOT A MYTH! It is a fact!

[minibubba]

It's a PR rating that AMD uses. Intel and AMD chips are designed very different and there is no point comparing them on a mhz for mhz basis. So AMD uses this PR rating to help people that are accustom to Intel's 'mhz is everything' approach. Otherwise people would look at AMD's 2.2 ghz 3200+ and then look at Intel's 3.2 ghz P4 and assume the AMD chip performs like a 2.2 P4 which would be a very false idea. So I don't really care how big the mhz gap is as long as the chip can perform. Granted the 3200+ can't quite catch up to the 3.2 P4, but it is close enough in my book.

[zeradul]

Megahertz and Gigahertz are IRRELEVANT measurements of a processor's capability.

It would be like selling a car with an engine that can go 16,000 RPM. Well what the f*ck does 16,000 rpm mean? Two engines from different brands both with 16,000 RPM might be WAY different in total power depending on how they are designed.

webe3, does a Celeron 2 Ghtz = P4 2 Ghtz???

[zeradul]

Let me clarify. The only time that Megahertz or Gigahertz IS Relevant is when you are comparing directly between the same processor line from the same brand, with the same core design, same transistor size, and same bus speed. Then it is SOMEWHAT relevant.

[webe3]

webe3, does a Celeron 2 Ghtz = P4 2 Ghtz??? [/B][/QUOTE]



From what I have read..... NO! I have heard that the celeron 2ghz is a value chip and aimed at lower end markets and does not have all the bells and whistles of a pentium 4 2 ghz. But I am no processor expert! To tell you the truth I have not compared the benchmarks of those two!

[webe3]

Let me clarify. The only time that Megahertz or Gigahertz IS Relevant is when you are comparing directly between the same processor line from the same brand, with the same core design, same transistor size, and same bus speed. Then it is SOMEWHAT relevant.

Hmmmm.......so I guess you are saying that I am comparing "apples to oranges"????

[zeradul]

Pretty much... ;)

This is how they stack up. http://www.tomshardware.com/cpu/2003...charts-23.html

[Wiggo]

Ok this is what happens, a P4 exacutes 6 instructions per cycle (IPC) where as pre Barton cores Athlons perform 9 IPC but since the Barton core release AMD has lost it's way with the PR ratin' as the Barton core still only carries out 9 IPC though they've lowered the MHz but upped the PR ratin' which I personally think is wrong as the extra L2 cache in most circumstances doesn't make it up which is likely why I stick to T'bred cores (the T'bred 2800+ was dropped b4 public sale as it'd still be a competitor to the 3200+ Barton).

[Woody_87th]

I'm a little more faithful than Wiggo on the value of L2 cache! :D It really depends on the applications you are running. Games do in fact benefit from larger L2 caches.

I will agree with Wiggo that I think AMD went a little too far in their number ratings to compare the Barton with the latest P4s. I don't think they had much choice from a marketing perspective but that doesn't mean the Barton is a poor CPU. I think the 3200 is probably in the performance area between a 3.06GHz P4 (533MHz FSB) amd the newer 3.0GHz P4 (800MHz FSB). There are so many other factors that affect a system's overall performance so this is of course not an absolute reference but seems to be typical with comparably equipped systems.

I also think the 3200+ is overpriced but again this is probably a wise move on AMD's part to keep the public under the perception that they have not fallen behind Intel. When the Athlon64 is released we may again see AMD take the lead but nothing is certain other than the fact that both companies will continue to improve their products.

Apple computer has proven that MHz is not as important as Intel would like you to believe with their systems that clock considerably slower than Intel systems but process information in a very different manner to achieve performance. Apple uses the Motorola CPU which in it's latest version is a 2GHz 64bit CPU running on a 1GHz FSB. Apple designs their high end systems as dual CPU designs. The first AMD64s will probably run at similar speeds.

I personally don't invest in the most expensive CPU on the market and prefer to buy a cheaper version that can be overclocked. As I pointed out on a previous thread you can purchase five Barton 2500's for the price of a single 3200 and my 2500+ runs very nicely at the specs of a stock 3200 (200x11). This gives me a very good value for the money.

[maximus7001]

Everybody assumes that p4 is so great. Notice how the pIII's perform different as well. In zeradul's link, the pIII 1.2/pc133 performs the same as the p4 1.5/pc800. And the pIII 1.4/pc133 outperforms the p4 1.8/pc800. It's for the same reason. Anyone have a guess at where my rig comes in on that chart?(assuming it has the same gfx card that they all had for that benchy)

[Woody_87th]

After making the above post I read the latest issue of MaximumPC where they tested three new 875 mobos with the 3.0 P4 and the exact same RAM and Vid card as me!! Interesting to see that my 2500+ kicked all their butts when clocked up to 3200+ default settings and all other settings standard including my vid card,

I don't have any 875 systems laying around to benchmark but I would tend to trust the benchmarks of such a magazine to be impartial and accurate. This means that the 3200+ on an NForce2 mobo is superior to a 3.0GHz P4 on any 875 mobo!!! :D

Also keep in mind that the 875 chipset is much newer than the NForce2 and the Bartons.

{Waits for all the Intel guys to throw old 500MB hard drives at me}