Submitted by Most current Intel and AMD CPUs, as well as almost every Macintosh model from the last few years,
geekbenched and compared. Interesting results if you just want to have an overall generic idea of how fast is what. Interesting to see that my 12″ 867 Mhz Powerbook G4 is at least 6 times slower than a new MacBook Pro or that new Core2Duo CPUs at 2 GHz are way faster than my pretty recent 3 GHz P4 (which came with hyperthreading and 64bit support nonetheless). Technology flies fast past you!
Performance Comparison of Most Current CPUs
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
You feel so slow sitting on a P4 workstation next to a Core 2 Duo gaming PC… 🙂 Would be nice to see some comparison to popular non-x86 etc. CPUs (e. g. Sun T1, DEC Alpha, IBM RS64 or MIPS). Are they completely obsoleted now?
Most people don’t have these… The linked article is about popular, and recent, CPUs.
I didn’t think that the Core2Duo at 2 GHz would be that much faster than one of the fastest versions of the 3 GHz P4 ever released (that I currently use as my main desktop)…
Yes, what about VIA C7 and C7D? They are modern and easily available. There is also a dual processor system in teir repertoir and PC2500 motherboards are about to hit the market. Intel/AMD dichotomy. It does not sound fair.
well, the P4 netburst arch is a POS. The core 2 is loooads better. So It’s not at all surprising to me.
Sun SPARC and IBM POWER are very much alive. From the numbers I’ve seen, IBM POWER6 will continue IBM’s silicon performance dominance by quite a bit. Perhaps a comparison of these two, and IA64, since they are all direct competitors?
I suggest you be reasonable. I don’t think that a web site like geekpatrol has access to all that hardware. Besides, most people, as I said, only care about normal cpus that they can buy off the shelf.
I’m the author of the article and the developer of Geekbench.
While I’d love to compare more hardware architectures, part of the problem is that I don’t have access to any “exotic” hardware (like SPARC, POWER, and IA64). Even with access, I’m not sure the effort of porting and maintaining Geekbench on these platforms would be worthwhile; I don’t know how many people on these platforms would be interested in Geekbench!
Nice results. Though, I think their benchmarks are more sensitive to memory bandwidth than is necessarily representative of the real world. I base this on the huge performance difference in their tests between G4 and G5 Macs. Yes G5’s are faster, but most real world benchmarks have indicated a smaller margin between the two, though everyone knows the G4’s biggest limits are the terrible memory/FSB speeds.
Several people have commented that Geekbench places too much emphasis on raw memory bandwidth (which certainly hamstrings processors like the PowerPC G4). It’s something I’m working on for the next version of Geekbench.
Its easy to see when comparing Athlons between socket 939 and AM2, which as far as I know the only difference is DDR vs DDR2.
If you want to see benchmark results for a wider range of CPU families, please have a look at
http://www.spec.org/benchmarks.html
.
Power 5+ has 5 out of the top 10 TPC boxes. 1 is a Power4+ box. 4 are Itanium.
http://www.tpc.org/tpcc/results/tpcc_perf_results.asp
(#5 is a Microsoft Windows Server 2003 Datacenter Ed.(64-bit)SP1 box)
But they are very expensive.
They have no showings on the Price/Performance chart.
http://www.tpc.org/tpcc/results/tpcc_price_perf_results.asp
(8 out of 10 are Windows)
Power6 is due in mid-2007.
Edited 2006-12-08 07:03
Power 5+ has 5 out of the top 10 TPC boxes. 1 is a Power4+ box. 4 are Itanium.
TPC-C is the last benchmark I would ever use as a comparison of CPU performance. Some companies do not even bother to submit result as the results of the benchmark are worthless.
TPC-C is the last benchmark I would ever use as a comparison of CPU performance. Some companies do not even bother to submit result as the results of the benchmark are worthless.
Why do you say that?
Beats me, but I’d like to know how in the heck Windows figures into price per performance or whatever of a CPU.
Some OS’s are better optimised for certain tasks than others with the same CPU, usually Solaris tops charts for being pretty quick with server applications. This is why they are included in the calculations.
I cannot speak for “drdoug”, but TPC-C is a very bandwidth constrained benchmark, and the memory controllers and chipsets are at least as important as the CPUs. The TPC-C benchmark is extremely server and DB oriented, and I suppose that most “geeks” are more interested in single-user experience.
TPC-C is the last benchmark I would ever use as a comparison of CPU performance. Some companies do not even bother to submit result as the results of the benchmark are worthless.
Why do you say that?
NotParker, TPC-C put aside the fact that TPC-C is very old (1992) with some very large holes which allow vendors scope to modify (legally cheat). The results of the benchmark are very dependant on many non-cpu factors. One of IBM’s results used 6400 hard disks to achieve the result. If they used more, they would have most likely got a better result.
TPC-C results only show how much spare time (for vendor optimsations) and spare hardware a vendor has. It also helps if you also produce the database software
NotParker, TPC-C put aside the fact that TPC-C is very old (1992) with some very large holes which allow vendors scope to modify (legally cheat). The results of the benchmark are very dependant on many non-cpu factors. One of IBM’s results used 6400 hard disks to achieve the result. If they used more, they would have most likely got a better result.
TPC-C results only show how much spare time (for vendor optimsations) and spare hardware a vendor has. It also helps if you also produce the database software
I understand that.
But what other way would you suggest to benchmark massive systems with 64/128/256 cores.
Or systems with 10’s of thousands of users.
It isn’t Geekbench.
Where else could I find out how much I would have to spend with D*** to service 53,000 users?
http://www.tpc.org/results/individual_results/Dell/Dell_Inc_98_0606…
But what other way would you suggest to benchmark massive systems with 64/128/256 cores.
Use a benchmark that is either based on your application or close to it.
DB OLTP – Wait for TPC-E
DB Data whorehouse – TPC-H
App server – SPECjbb
Web – SPECweb
etc…. Though always view there results as something close to speculation. It is the results a vendor does not submit is always more interesting
Or systems with 10’s of thousands of users.
It isn’t Geekbench.
Where else could I find out how much I would have to spend with D*** to service 53,000 users?
[/i]
Back to TPC-C. The benchmark was never designed (pre 1993) for 10,000’s of users. It consists of something like 5 simple transactions on nine tables. Anything more complex than a well used coffee shop, TPC-C would start losing its value.
For pure CPU benchmarking Geekbench is probably more use. What you are wanting is a complete systems benchmark. Not really what the article is about.
At first sight, Core 2 Duo looks like the most attractive CPU from a performance/watt point of view. I personally like PCs to be totally silent, and to not need a huge fan to keep it cool. So I’d get my hands on a Core 2 Duo if this machine here breaks down (if it ever will), and if I can resist my original AMD preference.
Still, most people and me too have budgets they have to take into account. Benchmarks are nice for the gamer with cash, but many others want affordable CPUs that are fast enough to run today’s bloatware nicely (and I’m including quite a few Linuxes here), and that are even reasonably power efficient. Well, let me not scare the reader with the average Dutch energy bill. Al Gore, too, would like us not to go straight for the P4 anymore.
So, some figures from a reliable, not that cheap computer store pricelist here in Rotterdam:
Intel Core 2 Duo E6300 (1.86GHz) [Geek P. rating 213.4] €209
Intel Pentium D 925 (3.00GHz) [Geek P. rating 192.1] €145
AMD Athlon 64 3000+ (Socket AM2 1.8Ghz) [Geek P. rating 117.8] €85
AMD Athlon 64 X2 3800+ (Socket AM2 2.00Ghz) [Geek P. rating 172.5] €179
The above Core 2 Duo 1.86Ghz has 81.15% more performance than Athlon 64 3000+ 1.8Ghz, at a 145% higher price, almost same frequency. This gives the single core Athlon 64 an advantage for many buyers. However, the same Intel compared to the dual core Athlon 64 X2 2.00Ghz has 23.7% better performance (and lower clockspeed) at 16.7% higher price. Here, the (admittedly smaller) advantage is to Intel.
The Pentium D 925 3.0Ghz has 63% better performance than Athlon 64 1.8Ghz at a 70% higher price. This makes it not a very attractive investment given the 1.2Ghz extra it needs to achieve this performance, producing heat and, I suspect, higher energy bills.
Vis à vis the Athlon 64 X2 2.0Ghz, the performance of the Pentium D 3.0Ghz is 11.3% better at only 80% the price. I wonder if that is enough to convince people who think of power consumption, and heat.
We might then compare a few Pentium 4 processors, and my guess is they will come out (in terms of, purely, performance/price) a little more favourably than Pentium D vs. AMD 64, but it’s not clear to me which models have the 64 bit extensions that should make these CPUs future proof, and which do not.
My conclusion would be that both Intel and AMD have been aware of the benchmarks, since the pricing is rather competitive. Performance per €/$/¥/.. is more or less comparable. AMD might have to adjust a few Athlon 64 X2 prices, and Pentium D is simply overpriced.
So I don’t think Core 2 Duo is the AMD killer so far, unless marketing makes it that. If AMD can manage to follow Intel in its 65nm (and smaller) path in time, it looks rather good in processor land, now that the dead end Ghz race seems sort of over.
BTW, it was fun to read that the G4 I have here looks rather miserable in the Geek Patrol list. It’s not fast, sure, but the graphics card makes up for it a little. Plus, it hardly makes any noise, and it hardly ever gets hot. I’m not a gamer/3D modeler anyway. So before the author actually does decide to artificially shorten his G4’s life span, please send it over to me and I’ll pay a few bucks for it too.
That said, this is only one benchmark. It’ll get shot at as most are.
“I personally like PCs to be totally silent, and to not need a huge fan to keep it cool.”
You’ll hate the most gaming PC coming from the shelf. 🙂
“The above Core 2 Duo 1.86Ghz has 81.15% more performance than Athlon 64 3000+ 1.8Ghz, at a 145% higher price, almost same frequency.”
Local dealer here in Magdeburg (Germany): Same CPU for 204,90 EUR on MB ASRock 775 Dual-VSTA for 59,90 EUR runs fine with NetBSD Live 2007. But the whole system makes too much noise.
“The Pentium D 925 3.0Ghz has 63% better performance than Athlon 64 1.8Ghz at a 70% higher price. This makes it not a very attractive investment given the 1.2Ghz extra it needs to achieve this performance, producing heat and, I suspect, higher energy bills.”
Surely, these settings are ideal for gamers who run the same system for not longer than a year. Most of them have their parents paying the bills. 🙂
>> “I personally like PCs to be totally silent, and to not need a huge fan to keep it cool.”
> You’ll hate the most gaming PC coming from the shelf. 🙂
I’m afraid I will.
If I’ll ever need another PC I’ll check out the most silent power supply + fan, and a silent CPU fan, after all power supplies are to blame for much of the noise.
But what can you reasonably do against a multi Ghz, dozens of Watts-eating monster like the P4?
>> “The Pentium D 925 3.0Ghz has 63% better performance than Athlon 64 1.8Ghz at a 70% higher price. This makes it not a very attractive investment given the 1.2Ghz extra it needs to achieve this performance, producing heat and, I suspect, higher energy bills.”
>Surely, these settings are ideal for gamers who run the same system for not longer than a year. Most of them have their parents paying the bills. 🙂
Not to mention the graphics card, and the 20+” CRT monitor that a hard core gamer “needs”.
http://www.silentpcreview.com/ is where I go for my silent computing needs… I’m currently sitting next to a A64 3200+ w/ a nVidia 6800, producing no noise what so ever.
none? do you have a flash based hdd?
that should produce a tiny bit of noise.
you should get 4 of those 4gb gigabyte sata flash drives and RAID them. be totally silent then
And faster…
Laptop HD: inside foam & suspended in the case (as not to produce vibrations.)
http://www.silentpcreview.com/ is where I go for my silent computing needs… I’m currently sitting next to a A64 3200+ w/ a nVidia 6800, producing no noise what so ever.
I’ve seen it, good website.
So.. out of curiosity, what CPU fan and power supply do the noiseless job for you?
AMD just released their 65 nm CPUs, and a 35W dual-core (albeit significantly more expensive) version – Athlon 64 3800+ EE – really is cool and quiet.
We use it exclusively here, and it’s fast and can run passively cooled with enough chassic airflow (read: 120 mm fan in the back at low speed).
It doesn’t beat the Core 2 Duo, which, admittedly, seem the better cpu these days, but being an AMD enthusiast, and running a 64-bit OS, the Athlon really is a nice enough alternative.
BeOS is very snappy on my dual P3 1Ghz box, thank you very much!
If you are looking for more benchmarks like this I would suggest Toms Hardware’s CPU charts here: http://www23.tomshardware.com/cpu.html
They have numbers from 50 or so processors and you can sort them by benchmark.
That’s an OK comparison, but I prefer the CPU Charts at Tom’s Hardware. Those charts list many seperate benchmark scores so shoppers can use the benchmarks that best reflect their target use. The Toms Hardware guide doesn’t include Macs though…
Also, the processors at Geek Patrol would be easier to compare if all of the processors were on a single chart, perhaps color coded for brand. Scrolling up and down across several different charts to compare processors is a PITA.
-Bob
added: [quote]BeOS is very snappy on my dual P3 1Ghz box, thank you very much![/quote]Yes, I’m sure that BeOS on a PIII will run faster and smoother than Vista on a Duo Core 2 Extreme too.
Edited 2006-12-08 16:37
now that Apple has switched to intel, the cpu wars are a lot less interesting as everything is all x86. Intel’s original plan was to get everyone to eventually switch to IA-64 but that changed because of AMD.
The good news with x86 as the standard is that it makes it a lot easier for developers and the AMD/Intel fight brings a lot of benefits.
On the other hand, it is ironic that all 3 major game consoles use IBM cpu’s. There’s less reason to get a faster cpu for playing games on a PC when you can play much better on an xbox 360. Intel and AMD will have to deal with that threat-something they never faced with the mac.