AMD Mobile Athlon XP 2600+ vs XP 25000+ ???
- Thread starter Jalapeno
- Start date
BlueWeasel
Lifer
- Jun 2, 2000
- 15,944
- 475
- 126
In theory, the 2600+ should OC better since it runs at a higher speed than the 2500+, but at the same voltage (1.45v). However, everything I've read so far seems to indicate the 2400, 2500, and 2600 seem to all hit about the same.
It all depends on the stepping you get.
It all depends on the stepping you get.
Just got my mobile 2600+ setup, and it's currently running Prime95 @ 2.5Ghz, 1.65v. I really didn't mind spending the extra $$ to get a chip that was almost running XP 3200+ speed @ 1.45v, and was unlocked. As an average, I'm sure the 2400+, 2500+, and 2600+ mobiles all o/c about the same level. However, in theory, the 2600+ will be the "best of the bunch".
Regardless, I'm happy with mine so far...I was really only expecting about 2.3 - 2.4Ghz.
Regardless, I'm happy with mine so far...I was really only expecting about 2.3 - 2.4Ghz.
Originally posted by: Insane3D
Just got my mobile 2600+ setup, and it's currently running Prime95 @ 2.5Ghz, 1.65v. I really didn't mind spending the extra $$ to get a chip that was almost running XP 3200+ speed @ 1.45v, and was unlocked. As an average, I'm sure the 2400+, 2500+, and 2600+ mobiles all o/c about the same level. However, in theory, the 2600+ will be the "best of the bunch".
Regardless, I'm happy with mine so far...I was really only expecting about 2.3 - 2.4Ghz.
Where did ya get your chip?
An XP 25000+ is going to totally whup on that lowly Mobile XP 2600+. I mean, after all, an XP 25000+ would be running at something like 18.3GHz, wouldn't it?
Originally posted by: Creig
An XP 25000+ is going to totally whup on that lowly Mobile XP 2600+. I mean, after all, an XP 25000+ would be running at something like 18.3GHz, wouldn't it?
yeah I think 18.3 Ghz sounds right hehe. I think it is mostly to do with how much money people want to spend. Most of the chips should get around the same OC since these are so mature.
the problem is finding a mother board that will support it, as even with a 22x multiplier and a 300MHz FSB, that only comes out ot 6.6GHz, which is easily only a 9000+ or so
Nah, you're all wrong. IPC on the K7 architecture wouldn't hold out that long... Even if AMD could somehow, magically get their K7 CPUs to run at 18.3GHz, it would take even more to get a 25000+ rating out of. Remember, IPC does not scale linearly with CPU speed increases/FSB speed increases (that's what new architectures are for). Turning the FSB up significantly would help, but there's really only so much memory bandwidth can do for the K7 architecture (it's not nearly as bandwidth-hungry as the P4). If you want a 25GHz CPU (effective) you'd need the following:
process change (die shrink + conductive material change/enhancement)
CPU would have to be extremely deeply pipelined (I'd say in the range of over 100 stages in each pipeline, just as a guesstimate compared to modern CPUs)
some fast effin RAM (2GHz FSB anyone?)
phase-change cooling (no way this CPU would run on air)
serious power tweakage/enhancements (no way would this thing run at 1.5V without *several* process changes down to the level of .03 micron and some "cool" die material, much "cooler" than what's currently being used)
There would be quite a bit involved in designing a CPU to run at 25GHz (effective) and I doubt it would be very much like a modern day CPU at all. Just to give you an idea, one could compare the late PII/early PIII to current P4C/E to get an idea of just what is involved here. In that timeframe the manufacturing process was shrunk from .25mu -> .18mu -> .13mu ->.09mu. That's 3 process changes with who knows how many thermal enhancements thrown in the mix to yield CPU speeds 6-7 times higher than what they were just a few short years ago. I would expect it to only get tougher from here on out so one could reasonably assume it would take *at least* another 3 process changes to get to 25GHz. Even if it only did take 3 process changes, Intel would have to get their process down to .03 micron (.25mu -> .09mu: 2.777 ratio, .09mu -> .03mu: same ratio).
anyway, that's just my two cents... I could keep going but I think you all get the idea.
process change (die shrink + conductive material change/enhancement)
CPU would have to be extremely deeply pipelined (I'd say in the range of over 100 stages in each pipeline, just as a guesstimate compared to modern CPUs)
some fast effin RAM (2GHz FSB anyone?)
phase-change cooling (no way this CPU would run on air)
serious power tweakage/enhancements (no way would this thing run at 1.5V without *several* process changes down to the level of .03 micron and some "cool" die material, much "cooler" than what's currently being used)
There would be quite a bit involved in designing a CPU to run at 25GHz (effective) and I doubt it would be very much like a modern day CPU at all. Just to give you an idea, one could compare the late PII/early PIII to current P4C/E to get an idea of just what is involved here. In that timeframe the manufacturing process was shrunk from .25mu -> .18mu -> .13mu ->.09mu. That's 3 process changes with who knows how many thermal enhancements thrown in the mix to yield CPU speeds 6-7 times higher than what they were just a few short years ago. I would expect it to only get tougher from here on out so one could reasonably assume it would take *at least* another 3 process changes to get to 25GHz. Even if it only did take 3 process changes, Intel would have to get their process down to .03 micron (.25mu -> .09mu: 2.777 ratio, .09mu -> .03mu: same ratio).
anyway, that's just my two cents... I could keep going but I think you all get the idea.
Originally posted by: Jalapeno
OK, OK - ENOUGH already! Probably, I was under the influence of too much :wine: :wine: :wine: :wine: :wine: :wine: :wine: :wine: when I wrote the post...
Don't make me edit it...
comeon, join in the fun and make fun of your typo with us! We could discuss the TDP of such a CPU or perhaps we could discuss how many functional units such a CPU would need to keep the extremely-deep pipeline fed at all times. Oooh! I know! Let's discuss the problem of getting SRAM to run at such high frequencies, perhaps limiting the amount of on-die cache that can be used or even preventing its use altogether without the aid of a frequency divider to run the cache at lower speeds.
LMK
- Dec 26, 2000
- 991
- 10
- 81
Originally posted by: Flyermax2k3
Originally posted by: Jalapeno
OK, OK - ENOUGH already! Probably, I was under the influence of too much :wine: :wine: :wine: :wine: :wine: :wine: :wine: :wine: when I wrote the post...
Don't make me edit it...
comeon, join in the fun and make fun of your typo with us! We could discuss the TDP of such a CPU or perhaps we could discuss how many functional units such a CPU would need to keep the extremely-deep pipeline fed at all times. Oooh! I know! Let's discuss the problem of getting SRAM to run at such high frequencies, perhaps limiting the amount of on-die cache that can be used or even preventing its use altogether without the aid of a frequency divider to run the cache at lower speeds.
LMK
But then it would be in order to move the thread to the "Highly Technical" forum, wouldn't it?
Go ahead, make fun. It's very interesting actually!hehe, not much else to say at this point without getting into serious architectural details and actually designing a "CPU of the future".
I'll leave that one up to AMD and Intel, if you don't mind Who knows what those guys have up their sleeves? Would you have been able to predict 3+ GHz clock frequencies, 800MHz/1000MHz FSBs, 64-bit extensions, HyperThreading (multiple logical CPUs on a die), SMP/multiple physical CPUs (on a single die), and a whole host of others only a few short years ago? I foresaw a few of them (AMD64 has been talked about since 2000, with initial rumors starting even before the original K7 Athlon came out) as well as multiple physical CPUs on a die but the clock frequencies and software extensions (SIMD, HT) hadn't really ever entered my mind until around the time they were coming out anyway.
All I'm getting at is who knows where we'll be in another 5 years? Intel claims frequencies of 10GHz+ within the next 2-3 years, and both companies have plans for multiple physical AND logical CPUs on a single die (i.e. 2/4/8 CPUs in one!) but I don't know what other improvements will be made to the world of x86/consumer CPUs in that timeframe. One can logically assume that more SIMD extensions will be added (there's always some feature developers are screaming for). One tradeoff Intel/AMD may have to make to be able to scale their CPUs so high is in IPC (instructions per cycle). The P4 (especially the Willamette) is a perfect example of this. Going from the PIII to the P4, Intel had to sacrifice IPC to be able to scale their CPUs up to extremely high clock speeds (in comparison with the PIII). When designing a CPU, one has to make tradeoffs between IPC and clock speed. AMD chose IPC, Intel chose clock speed. We've been seeing the results of those decisions for the last 3-4 years now.
Interesting times ahead
I'll leave that one up to AMD and Intel, if you don't mind
Who knows what those guys have up their sleeves? Would you have been able to predict 3+ GHz clock frequencies, 800MHz/1000MHz FSBs, 64-bit extensions, HyperThreading (multiple logical CPUs on a die), SMP/multiple physical CPUs (on a single die), and a whole host of others only a few short years ago? I foresaw a few of them (AMD64 has been talked about since 2000, with initial rumors starting even before the original K7 Athlon came out) as well as multiple physical CPUs on a die but the clock frequencies and software extensions (SIMD, HT) hadn't really ever entered my mind until around the time they were coming out anyway. All I'm getting at is who knows where we'll be in another 5 years? Intel claims frequencies of 10GHz+ within the next 2-3 years, and both companies have plans for multiple physical AND logical CPUs on a single die (i.e. 2/4/8 CPUs in one!) but I don't know what other improvements will be made to the world of x86/consumer CPUs in that timeframe. One can logically assume that more SIMD extensions will be added (there's always some feature developers are screaming for). One tradeoff Intel/AMD may have to make to be able to scale their CPUs so high is in IPC (instructions per cycle). The P4 (especially the Willamette) is a perfect example of this. Going from the PIII to the P4, Intel had to sacrifice IPC to be able to scale their CPUs up to extremely high clock speeds (in comparison with the PIII). When designing a CPU, one has to make tradeoffs between IPC and clock speed. AMD chose IPC, Intel chose clock speed. We've been seeing the results of those decisions for the last 3-4 years now.
Interesting times ahead
Originally posted by: screw3d
Originally posted by: Insane3D
Just got my mobile 2600+ setup, and it's currently running Prime95 @ 2.5Ghz, 1.65v. I really didn't mind spending the extra $$ to get a chip that was almost running XP 3200+ speed @ 1.45v, and was unlocked. As an average, I'm sure the 2400+, 2500+, and 2600+ mobiles all o/c about the same level. However, in theory, the 2600+ will be the "best of the bunch".
Regardless, I'm happy with mine so far...I was really only expecting about 2.3 - 2.4Ghz.
Where did ya get your chip?
Sorry for the delay in my reply...
I got it @ Newegg.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
-
-
Facebook
Twitter