AMD 64 3000+ or P4 3.2ghz HT?
- Thread starter FX5900
- Start date
This is why people think the A64 runs cooler (at both load AND idle) than even a P4 NW. I don't think this is the only graph that demonstrates AMD's superiority in terms of power draw (and, thus, heat and noise creation).
Judging by the date of that review, those A64s are likely 130nm Clawhammer, not 90nm Winchester, cores. Winchesters should drop power draw even lower, whereas I don't think even the latest 90nm Prescotts can compete with Northwoods in terms of power draw.
Edit: This ZDNet UK review shows the latest P4 6xx-series is pretty close to a 130nm A64 in terms of power draw. But this AT graph is a bit more interesting. The NW-based P4EE compares favorably to a 130nm A64 in terms of idle draw, but check out how the 90nm A64 compares to the 130nm one and the NW. Most comparisons show AMD is in the lead WRT power draw.
I'm really curious to see how Intel's dual-core P-D compares to AMD's dual-core. Hopefully we'll find out around April 21st.
Judging by the date of that review, those A64s are likely 130nm Clawhammer, not 90nm Winchester, cores. Winchesters should drop power draw even lower, whereas I don't think even the latest 90nm Prescotts can compete with Northwoods in terms of power draw.
Edit: This ZDNet UK review shows the latest P4 6xx-series is pretty close to a 130nm A64 in terms of power draw. But this AT graph is a bit more interesting. The NW-based P4EE compares favorably to a 130nm A64 in terms of idle draw, but check out how the 90nm A64 compares to the 130nm one and the NW. Most comparisons show AMD is in the lead WRT power draw.
I'm really curious to see how Intel's dual-core P-D compares to AMD's dual-core. Hopefully we'll find out around April 21st.
ninelven said:
A winchester runs an average of 2-5 degrees less so just subtract that. Either way the two are relatively close to each other.
I agree A64s run cooler and draw less power compared to Prescotts (though less so with the new 6xx), but not Northwoods. Yes they fair better in games but I personally don't notice a difference between 70 and 80 fps and P4s are still the way to go for really anything besides gaming atm.
The max TDP that Intel quote for their chips isn't actually accurate - max consumption is actually a bit higher:
http://www.cpuheat.wz.cz/html/Pentium4.txt
Max heat dissipation of a 3.2Ghz Northwood is, according to these measurements around 96W. By comparison, the max TDP of the current Athlon 64 line (up to and including the fastest FX chips on the 130nm process) is just 89W. I think it's safe to say that the lower-speed A64 3000+ on the 90nm process will dissipate considerably less heat than this and as such are much cooler than Northwood (cooling-wise Northwood is still a better option than Prescott if you do go the Intel route though).
I think the expected max TDP for the new dual-core A64s is to be 109W (not sure where I remember that figure from). This will probably include the 2.6Ghz dual-core if memory serves so the slower speed dual cores should dissipate much less than this. For comparison, the first releases of the new Pentium D reputedly have max TDP of around 130W.
The new Revision 'E' Venice chips will run a little hotter than Winchester. However, the strained silicon used in them allows higher clocks and, in theory, lower voltages at the same clocks as can be seen in the forthcoming Turion chips based on the same (or at least very similar) technology. Venice chips are expected to become available later this month and will replace Winchester as production ramps up (I think).
http://www.cpuheat.wz.cz/html/Pentium4.txt
Max heat dissipation of a 3.2Ghz Northwood is, according to these measurements around 96W. By comparison, the max TDP of the current Athlon 64 line (up to and including the fastest FX chips on the 130nm process) is just 89W. I think it's safe to say that the lower-speed A64 3000+ on the 90nm process will dissipate considerably less heat than this and as such are much cooler than Northwood (cooling-wise Northwood is still a better option than Prescott if you do go the Intel route though).
I think the expected max TDP for the new dual-core A64s is to be 109W (not sure where I remember that figure from). This will probably include the 2.6Ghz dual-core if memory serves so the slower speed dual cores should dissipate much less than this. For comparison, the first releases of the new Pentium D reputedly have max TDP of around 130W.
The new Revision 'E' Venice chips will run a little hotter than Winchester. However, the strained silicon used in them allows higher clocks and, in theory, lower voltages at the same clocks as can be seen in the forthcoming Turion chips based on the same (or at least very similar) technology. Venice chips are expected to become available later this month and will replace Winchester as production ramps up (I think).
ANova said:ninelven said:
A winchester runs an average of 2-5 degrees less so just subtract that. Either way the two are relatively close to each other.
I agree A64s run cooler and draw less power compared to Prescotts (though less so with the new 6xx), but not Northwoods. Yes they fair better in games but I personally don't notice a difference between 70 and 80 fps and P4s are still the way to go for really anything besides gaming atm.
DudeMiester
Regular
Actually P4s are only insignificantly faster in media encoding/decoding, and a few rendering programs. However, in business programs (word, excel, etc), compilers, servers, games, and basically everything other then media, athlon64s are faster. Furthermore they are cooler and use a lot less power. You have to remember Intel underestimates their TDP and AMD overestimates, and still AMD's are much lower. In real world tests an A64 Winchester will use about 85W, and a P4 about 115W. Big difference, although the temps may be similar because the P4's heatsink is 2x as big.
Also, for air cooling you can get more performance from overclocking the Athlon64's then you can from the P4's. Mainly because the P4s run hot. Oh and don't forget the countless other major features of the Athlon64 that the P4s utterly lack. First, an integrated very high performance memory controller, 64-bit operation that is actually faster then 32-bit rather then slower, a non-swamped FSB due to no memory usage and HTT, a better branch predictor, and all it takes is a BIOS update to support dual-cores instead of buying a whole new comp.
Basically, AMD has the best products out right now in all areas. They made Intel look like a bunch of idiots, although I admit the only idiots at Intel are in the marketing department. Unfortunitly for Intel, the marketing department runs the whole damn company, a fatal mistake.
Also, for air cooling you can get more performance from overclocking the Athlon64's then you can from the P4's. Mainly because the P4s run hot. Oh and don't forget the countless other major features of the Athlon64 that the P4s utterly lack. First, an integrated very high performance memory controller, 64-bit operation that is actually faster then 32-bit rather then slower, a non-swamped FSB due to no memory usage and HTT, a better branch predictor, and all it takes is a BIOS update to support dual-cores instead of buying a whole new comp.
Basically, AMD has the best products out right now in all areas. They made Intel look like a bunch of idiots, although I admit the only idiots at Intel are in the marketing department. Unfortunitly for Intel, the marketing department runs the whole damn company, a fatal mistake.
These threads take up approximately ~30% of the Internet (which is a good thing for those looking to sell CPUs). I think the online consensus is very distorted from reality, but then I would think that, being on here myself. There is nothing wrong with the Pentium 4. Likewise I could report to you that there is nothing wrong with the Athlon 64. Why would two established IC manufacturers base their business on something that is fundamentally flawed?
You will not be disappointed no matter which chip you buy...or...you will be disappointed in whatever chip you buy, because you always want more (which is normal and understandable).
The Pentium 4 is a new type of design and therefore it suffers from certain market conditions which may appear as flaws. The Athlon 64 is more tradtitional in its approach and handles legacy applications much better, or should I say more efficiently, and is great too. What you have to ask yourself is what you will be spending most of your time telling your CPU what to do. You also have to ask when you want the performance benefit. This latter point I want to emphasize because it can only grow in NetBurst's (Pentium 4) favor unless everyone suddenly decides that x87 was a really great a idea and should never be forgotten (stranger fundamentalist movements have taken form).
I think one of the stron parts of Pentium 4 is that it does really well in multithreaded environments thanks to HyperThreading. It also does really well in some specialized benchamrks because this is how it was designed, not because the benchmarks are biased. Athlon 64 looks bad in some benchmarks but does really well when the proverbial excrement hits the fan; it has a certain toughness that the NetBurst design lacks (on purpose).
The catch with Athlon in online discussions seems to be that Athlon 64 does really well in games. That's great and, let's face it, games sell. But games have a limited benefit from CPU power. Games are played in real-time and there is a certain level of performance that will please most people. Benchmarks have a tendency to drag you in to the whole "bigger is better" mentality, but really, at some point you have enough FPS or the bottleneck is elsewhere. That is not necessarily always so with some other compute intensive apps like video and audio encoding. In those circumstances you want all the speed you can get because these task consume all of your PC and it must finish before you can get on with other things. Luckily, that is not always so because HyperThreading can squeeze in secondary code while compute intensive tasks are running, allowing you to browse the Web comfortably while encoding video, for example. This is not true with the Athlon 64 design, which likes to dedicate all of itself to one task (and does so really well).
I am currently using an Athlon 64 as my main workstation. This is because I had never owned an AMD part and I thought it was painfully silly for someone like myself (having been computer addicted since 1982) . I like it. It's fast and has a knack at performing a lot better than specific benchmarks would suggest, but I still miss HyperThreading from the Pentium 4. Luckily, AMD is not so adamant about making money on every little switch in technology so I can probably keep using this same Socket 939 board when dual core is available and that should be a major step up from HyperThreading.
If you take away anything from this, I hope it is that many online presentations of the opposing technologies are very exaggerated and that the reality with most gaming is that we are still bottlenecked by the video card (at least in FPS games).
You will not be disappointed no matter which chip you buy...or...you will be disappointed in whatever chip you buy, because you always want more (which is normal and understandable).
The Pentium 4 is a new type of design and therefore it suffers from certain market conditions which may appear as flaws. The Athlon 64 is more tradtitional in its approach and handles legacy applications much better, or should I say more efficiently, and is great too. What you have to ask yourself is what you will be spending most of your time telling your CPU what to do. You also have to ask when you want the performance benefit. This latter point I want to emphasize because it can only grow in NetBurst's (Pentium 4) favor unless everyone suddenly decides that x87 was a really great a idea and should never be forgotten (stranger fundamentalist movements have taken form).
I think one of the stron parts of Pentium 4 is that it does really well in multithreaded environments thanks to HyperThreading. It also does really well in some specialized benchamrks because this is how it was designed, not because the benchmarks are biased. Athlon 64 looks bad in some benchmarks but does really well when the proverbial excrement hits the fan; it has a certain toughness that the NetBurst design lacks (on purpose).
The catch with Athlon in online discussions seems to be that Athlon 64 does really well in games. That's great and, let's face it, games sell. But games have a limited benefit from CPU power. Games are played in real-time and there is a certain level of performance that will please most people. Benchmarks have a tendency to drag you in to the whole "bigger is better" mentality, but really, at some point you have enough FPS or the bottleneck is elsewhere. That is not necessarily always so with some other compute intensive apps like video and audio encoding. In those circumstances you want all the speed you can get because these task consume all of your PC and it must finish before you can get on with other things. Luckily, that is not always so because HyperThreading can squeeze in secondary code while compute intensive tasks are running, allowing you to browse the Web comfortably while encoding video, for example. This is not true with the Athlon 64 design, which likes to dedicate all of itself to one task (and does so really well).
I am currently using an Athlon 64 as my main workstation. This is because I had never owned an AMD part and I thought it was painfully silly for someone like myself (having been computer addicted since 1982) . I like it. It's fast and has a knack at performing a lot better than specific benchmarks would suggest, but I still miss HyperThreading from the Pentium 4. Luckily, AMD is not so adamant about making money on every little switch in technology so I can probably keep using this same Socket 939 board when dual core is available and that should be a major step up from HyperThreading.
If you take away anything from this, I hope it is that many online presentations of the opposing technologies are very exaggerated and that the reality with most gaming is that we are still bottlenecked by the video card (at least in FPS games).
I find it interesting that people keep glossing over the fact that you'll be able to upgrade to a dual-core A64 with merely a BIOS update and a new CPU, whereas you'll need a new MB for a dual-core Pentium. So, even if the P4 is overall comparable in speed, power, heat, and (least of all) price, the A64 still has a far simpler upgrade path.
People arguing in message boards over these relatively minor details tend to have more time than money, so the A64's simpler upgrade path would appear to be rather important.
Price is the deal breaker. From NewEgg:
P4 3GHz 512kB OEM $201
P4 3.2GHz 512kB OEM $257
A64 3000+ 512kB RTL $146
A64 3200+ 512kB RTL $190
All signs point to A64 for the home user, from where I'm standing. But some people may value HT's purported greater "smoothness" with multiple apps and superior handling of streaming data, at least until dual-core CPUs become available later this year. In that case, I guess an extra $50-100 up front may be worth it.
Otherwise, the 3200+ compares pretty well to the P4 3GHz at the $200 price point. It's 10-15% slower in most video encoding, and 20% faster in UT2K4, Far Cry, and Doom III. I'd take the higher performance in the real-time app.
People arguing in message boards over these relatively minor details tend to have more time than money, so the A64's simpler upgrade path would appear to be rather important.
Price is the deal breaker. From NewEgg:
P4 3GHz 512kB OEM $201
P4 3.2GHz 512kB OEM $257
A64 3000+ 512kB RTL $146
A64 3200+ 512kB RTL $190
All signs point to A64 for the home user, from where I'm standing. But some people may value HT's purported greater "smoothness" with multiple apps and superior handling of streaming data, at least until dual-core CPUs become available later this year. In that case, I guess an extra $50-100 up front may be worth it.
Otherwise, the 3200+ compares pretty well to the P4 3GHz at the $200 price point. It's 10-15% slower in most video encoding, and 20% faster in UT2K4, Far Cry, and Doom III. I'd take the higher performance in the real-time app.
Pete said:
Jisted for truth. This is a real deal maker/breaker (or should be).
For the ethusiast who doesn't want to replace the mainboard, AMD has been much kinder historically and hopefully they will continue to be so. Of course there are points where the socket must be switched so that time is sensitive and equalizes the playing field. I think Socket 939 still has some life left (DDR-400, etc, etc) so it is a nice platform to buy into.
If you are buying Dell (nothing wrong with that IMO) then you really shouldn't care because you get everything new, but as a dabbler you need/want the flexibility to upgrade using the same components as a base. A definite plus in AMD's favor.
i own a p4 and im very disappointed with it. performance in ANY game that is cpu hungry fucking sucks. no if ands or buts about it.
and theres a HUGELY noticeable diff between a p4 and a64 in gaming. 15 to 30 fps is extremely noticeable unless your just a newbie who has no idea how to play. i could rly care less if my videos encode 1 minute faster, the a64 stomps anything intel has in almost every other category for a cheaper price, lower power draw, and a cooler system.
and theres a HUGELY noticeable diff between a p4 and a64 in gaming. 15 to 30 fps is extremely noticeable unless your just a newbie who has no idea how to play. i could rly care less if my videos encode 1 minute faster, the a64 stomps anything intel has in almost every other category for a cheaper price, lower power draw, and a cooler system.
ANova said:
K8 is a really new architecture whereas P4 is not as advanced, and was primarily designed to be able to reach high frequencies (marchitecture).
the opteron are a lot more efficient in dual processor configuration than intel parts, the difference increases even more when you increase the number of CPU, thanks to an architecture thought from the start for multi-processors/multi-core.
and 64 bits has a real added value (not only adress space, but increased number and size of registers), NX bit too. power consumption and heat make it more suited to build powerful server that uses a minimum of space.
opteron just makes a better server platform. i choosed for our critical servers (os being linux 64 bits), and am very glad with our investment.
in x86 world intel is just playing catch-up with AMD, implementing what AMD creates, just a lot later, with the notable exception of SSE2/3..
besides SS3, for what is the P4 better than AMD 64 bits ?
ninelven said:
Well actually, the figures I quoted show that the Northwood is significantly hotter than the equivalent Winchester cores. The max TDP for the current A64 line up to and including the fastest FX chips running at high clocks on the 130nm process is still less than for the 3.2Ghz Northwood.
Therefore the lower clocked 90nm Winchesters run much cooler than the Northwood equivalents. Luckily, the Northwoods are still reasonably cool in comparison to the Prescotts. Unless you're really interested in silencing your PC, a Northwood is fine in a normal PC.
As other people have mentioned in this thread, the Winchesters are faster than Northwood equivalents in some things and slower in others. However, they are also cheaper, cooler, support 64-bit instructions (bear in mind that MS have finally officially released Win64) and support a direct upgrade path to dual-core chips requiring nothing more than a bios update! I can't see why anyone would want to go for the Northwood unless you were specifically buying a PC for video editing. Also bear in mind that Winchester chips overclock very well indeed without too much effort if that's the way you want to go.
