Being in the corporate environment of Wall Street that has been heavily in bed with Sun, I have been trying to get my corporate IT to shift focus from Sun Sparc Servers to Sun AMD Operton Servers. I am getting alot of resistance from the Sun Sas because tests that we have done that have been Solaris vs Dell Linux boxes haven't been Apples to Apples. Sun itself won't do direct comparisons between their Sparc and x86 platform. I would love to see an article aimed at Sun Sparc Solaris vs Sun Operton Solaris vs Sun Operton Linux vs Dell Intel Linux that focuses on business type applications (databases, web serving, mathmatical libraries, etc).
I'm planning on building a new workstation PC, primarily for use with solidworks. I'd love to see some wide comparisons using the workstation benches of single and Dual core Athlon and Pentium systems head-to-head, as well as some video card comparos using workstation class and consumer class cards, to find out where the hard earned dollars are best spent. Ram configurations could also be useful (512 vs 1GB vs 2GB), to see where optimal conditions are.
How about some 64 bit benchmarks using the Microsoft beta's. Is that not allowed? It seems like we have seen very little (other than your excellent recent server article) that compares 64 bit performance. And what little there is only looks at Linux. Microsoft made some vague claims on a webcast (64 bit way faster), but I'd love to see some Visual studio 64 bit compiler or SQL Server benches, AMD vs. Intel.
Though I probably couldn't recognize ActiveX if I saw it, the comments work perfectly fine for me with Opera, on Linux. All it seems to be using is javascript. (Which I likewise would like to see done without, so I can middle click it to open in a background tab, which is unpossible with javascript.)
Anand, how about a benchmark of a dual-P-M server compared to a dual Opteron and a dual Xeon server? A P-M provides very interesting alternatives in the SFF/low noise desktop space, but how about blade servers or even regular servers? If it is technically possible to get benchmarks for a dual P-M, wouldn't it also be similar to the upcoming dual-core P-Ms?
Alternately, how about a fanless P-M desktop? It would be interesting to know if it would stand up to the rigors of heavy workloads.
I second the Quicktime 7 suggestion of Eric Florenzano. Actually what I think Eric means, and what I mean, is that it would be great to see a review that uses an H.264 video clip as a stress test. Although quicktime 7 is the big name I'm not sure if it's the best for benchmarking.
I would like to see some comparison of the heat output and noise of these new cards. CPU's have had issues with too much heat and power consumption, about time it hit graphics cards in a big way?
here's an idea, would you by any chance have a beta or something of the unreal 3.0 engine. if it really has multithreading enabled, it would be nice to see how the X2 or PD/EE compares with single core FX's P4/EE's.
Personally, I'd love to see some sort of benchmarks on chess positional analysis software such as chessbase 9 (with deep fritz 8 as the analysis engine.) I don't know how doable this would be or if there is enough demand to make it worthwhile, but I do quite a bit of this and find it to be very cpu-intensive, since it uses a "brute-force" method of anlysis.
I'd like the "comments" link not be ActiveX activated. I block ActiveX as much as I can to protect vs embedded spyware and if there's no good reason for you to use ActiveX, maybe you could consider letting it go.
I would like to see some reviews of the Apple Cinema 30 inch display running on a PC. I know that it requires a dual dvi connection and not too many cards outside of workstations provide that.
Only suggestion would be to see what the single and duel core chips do in multitasking on something other than Windows (Linux/FreeBSD or some such). I'm a bit curious as to how the windows scheduler affects multitasking performance (concidering the rather common ideas that the scheduler is crap). Having "virtual" processors through HT might alleviate some of the inherant issues in the schduler (as seen on various Intel vs AMD benches measuring HT impacts). I don't know if this has been done before, but it'd be an interesting thing to look into.
#44: That would be useless. First of all, both of the OSes are pretty old and even though no one has conducted the benchmarks, people aren't looking for that anymore.
Secondly, there shouldn't be too much, if any, performance difference between the two OSes.
I agree with reference computers. I would belive the average life of a computer for someone that reads Anandtech is 2 or 3 years maybe? Having a 2-3yr old CPU on all the benchies would be sweet. Most people don't upgrade from a 3200+ to a 3400+...
It would be cool if you could build a flash widget to do a weighted comparo for processors.
Display all of the categories you have benchies in and what % the CPU is better than some other CPU. Then the user could tweak the sliders to add up to 100%, for what percent of usesage they use. So if someone only did office stuff they could tweak the slider to 100% office benchmarks and see much better CPU A say 10% is than CPU B. Wehere someone else mught tweak 10%office and 90% graphics and CPU A is only 5% better...Some sort of weighted average...
I'd like to see some reviews of raid 5 SATA. Plus i'd like to see those reviews as whats supported by Novell Netware. Sata is getting much more inexpensive and now with Sata II, it should start putting a hurting on SCSI or does it even come close?
-Gentoo compile time benchmark.
(default packages for a stage-1 install, scripted total compile time).
-More Reference CPUs included in benchmarks.
Perhaps one representative CPU at each milestone in common/recent CPU evolution; P3 1.0Ghz, P4 3.2Ghz Northwood, P4 Prescott, Pentium-M 1.6, etc(and AMD, of course).
thursday, nvidia g70 7800gtx launch, just a guess.
Consider reviewing Gigabyte's Ram-disk,(pci slot ddr-ram hard drive) I think this could load games very fast and work well with video editing.
Also, the Antec P180 is an interesting new case that has many good features at a reasonable price. thanks
Imo, we dont see enough quake III based benchmarks anymore, even though some of the ore modern games, such as Jedi Knight Jedi Academy, are still widely played. Why not have a high res benchmark of something in quake 3 to see whether CPU impacts performance or not, after all, we dont buy an FX57 (im assuming) or whatever near $1000 dollar CPU this is in order to run synthetic benchmarks or games at low res, etc... Maybe you could save some people some money by pointing out in games at higher resolutions at what point a faster CPU becomes useless (3000+? 3500+? 1 meg of cache needed?) and its all graphics limited.
Interesting. The anticipation exhibited by some people at the announcement of a soon to be released article becomes too great if the soon does not equal tomorrow. Maybe an email alert system upon completion of the article would help. Maybe not giving out information on up-and-coming articles too far in advance would help as well.
It seems to be a similar situation to the HL2 release – many people anxiously waiting and perceived lack of information or unexpected delays turn into hostility towards Valve.
I would like to see more price/performance numbers and overall summary charts of all benchmarks in one graph. Something like xbitlabs has showing one CPU as the 100% baseline and then showing how much over and under the "next speed bump CPU" is compared to the baseline. Looking at pages and pages of benchmarks can be a hassle. Sometimes a table of percentages (Anandtech has used this in past) and a graph of all benchmarks on the x-axis and percents on the y-axis can be very helpful.
With the dramatic increase of AMD boards mostly using Nvidia chipsets, how about an article on the Nvidia chipset functions -firewall, active armor etc.- how to set up the firewall, what the settings mean, and how they affect real-world use like surfing, online gaming and even P2P. (Most of that set-up stuff is incomprehensible to "normal" folk)
Then run benchmarks on the different setups.
How about your own MS Office based benchmark?
Like loading a 20,000 page word doc.
How about distributed computing? Are there any of them still left that have work units that are always the same?
How about PDF rendering? I know some of the large ones take a rediculusly long time to load.
What about desktop searching? I know its proabably mostly a function of the disk speed but I wonder if CPU makes a difference how fast Copernic or Google desktop search indexes your disk...
Why not write your own C# program that does something as simple as adding 1+1 100million times... is C# optimized for AMD/Intel...
One thing I'd like to see in CPU gaming benchmarks is minimum framerates as in GPU benchmarks, rather than just averages.
It would be interesting to see how Intel/AMD platforms affect minimum framerates, and whether these results differ from average framerates or not.
I wouldn't mind seeing some performance comparisons among recent CPUs. You mention faster single core vs. slower dual core, I think would be great across both X2 and Pentium D, a "mini-roundup" of a bunch of the latest CPUs would be very useful.
I'd like to pick up on what ViRGE said. I'd love to see some benchmarks of actual power consumption on various systems "as a whole", rather than just looking at the heat generation or performance of the processors. How many watts are consumed under various use scenerios by competing systems and what percentage of the total is attributable to the individual components? For those of us running our systems 24/7, the energy costs add up.
I'd like to see more Photoshop benchmarks as that is what I am interested in. Also Photoshop is interesting as some things in it are single threaded and some are multithreaded.
Here's an idea(and it may just be sleepiness talking): create a suite of common, low-impact applications that would represent typical computer usage(say something similar to PC WorldBench), but instead of measuring time as a way to see raw performance, use it as a chance to measure other system attributes. For example, if I'm just surfing the web while listening to music and writing a term paper, what kind of power draw, noise, CPU usage, and heat output be? A lot of tests you do are all about performance at full speed, but when I'm fumbling around with only minor CPU usage, would a dual CPU system be quieter? A faster CPU(lower percentage of CPU time needed to do the task) versus a slower CPU(doesn't need to be tuned for high-clock, high-power usage applications)? I'm sort of envisioning something similar to previous Cool & Quiet / EIST tests, but with a focus on only partial CPU usage, instead of full/idle testing.
hmmm...NDAs expire on Thursday... Gnomedex starts Thursday...any connection? any announcements coming at Gnomedex, technology wise. I have heard though that Microsoft plans to announce some big things at the conference but I thought that was RSS related...
anyway...let's see!
(maybe they'll announce that the Xbox will have a PPU now :) )
or AMD has switched to Unified Shaders...or ATI has given up Unified Shaders...:)
I know you already run compiling benchmarks with Visual Studio, but I'd like to see a benchmark aimed at web developers -- something in PHP or ASP.NET, using MySQL or MS-SQL. Many small web development shops use their own personal machines for development work, and then once something is done and tested they upload it to the server. So the benchmark could be something simple like an ASP.NET script that makes a bunch of calls to a database, with both IIS or Apache and the database running on the same machine where the web browser is loading the page.... and just see how long it takes to run. Of course, it might end up that the disk or something other than the CPU is the limiting factor, but it'd be interesting to find out. Obviously you're not testing concurrency like you would for a server test, but you're still putting a pseudo-server load on a desktop machine.
I agree with Demo,
Battlefield 2 (demo or the real deal, depending on the scheduled release of your article) benchmarks would be terrific. It would be really great, especially if you throw in some SLI comparisons or even some memory speed comparisons. This game is one of the more demanding games on my system that I have played, and will certainly prove to be a great benchmarking standard for many reviews well into the future. Those are my two cents.
- Creathir
I'd like to see some more compiling benchmarks, particularly .net compilers. Most Microsoft shops use C# or VB.net rather than VC++, which is what you're currently using in your development benchmarks.
If I can sneak in "Any requests" for the whole site:
Please, please - May we have the SFF roundup you were working on in March and hoped to have up in a few weeks?
I'd love to see more Macintosh articles including perhaps benchmarks and reviews of some of Apple's professional lines of software including the Final Cut Studio series of software.
Where THE HELL, is the 360 & PS3 article, wow, this is just RIDICULOUS, this is totally PATHETIC and I just CAN'T believe it.
Again, he gives an update on the article and what now? Another month waiting coming up next?
C'mon Anand, you are better than that, it's also a little hypocritical to just ignore what MANY people are interested about and start talking about the "Yet ANOTHER CPU review"--Please Anand, at least the 360 and PS3 article is something more original, something you don't see everyday, new CPUs come extremely often, I think you know what I mean. Let's do something different, this opportunities don't come very often, so let's get to work man, every day that passes by our (at least MINE) expectations go one step higher, you may as well dissapoint--It would be a first....
Thanks and I hope to see the article.....as soon as possible.
How about a look at how amount of RAM affects the Benchmarks. Many people might own this CPU, enthusiasts will load it with RAM (1 - 2GB), System Builders might not (512MB). Would be interesting to see the difference what with memory controllers, hyper this, hyper that...
Do large amounts of RAM still offer the amazing performance gains on the same CPU that they were supposed to back in the day or are we all about Video Cards, CPU power and money spending these days?
When I read the title of this piece I thought it was a general "Any requests?" for the whole site, and I would like to make a request that has nothing to do with CPUs:
- AMD motherboards, socket 939 PCI-E roundup.
Please... it seems AT reviewd the DFI nF4, was happy enough with it and left it at that. But the DFI is a nightmare for a lot of people, it would be nice (and fair!) to see how the other boards compare for those of us who are not so lucky with DFI.
A few microbenchmarks would be nice. The memory latency and bandwidth benchmarks from LMbench, the STREAM benchmarks, etc.
To explore the dual-core performance issues, consider the OpenMP NAS Parallel Benchmarks (http://www.nas.nasa.gov/Software/NPB/) and the HPCC benchmarks (http://icl.cs.utk.edu/hpcc/). Although "batch"-style benchmarks, they show quite some difference between the dual-core systems.
Well, overclocking, for one ;)
Other than that... compiling is mainly all I actually use it for, which you already have. How well it scales on multiple cores/CPUs would be nice to see, though. (In previous reviews, either whatever was compiled didn't support parallel compilation (make -jN), or it just wasn't enabled.)
I would like to see some BF2 cpu marks. I dunno if that is really fitting into what your wanting but it could put some strain on multitasking. I know it uses a bunch of ram but not so sure on CPU load.
I'd like to see a benchmark of the Quicktime 7 Beta for Windows, seeing as it's so CPU-Intensive, and I have yet to get this working smoothly with 1080p on my A64 3800+. It'd be interesting to know if the FX-57 could handle this better...Oops, I mean, the "next speed-bump CPU review."
i would like to see the bench of real media producer (i asked it few times already), and a new Divx 6 codec, but not normal version, but a DivX Helium version: http://labs.divx.com/archives/000055.html (that one is optimized for dual-core cpus)
well thats all, i find more interesting dual-core cpus anyway...
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
66 Comments
Back to Article
Fester - Thursday, June 23, 2005 - link
Being in the corporate environment of Wall Street that has been heavily in bed with Sun, I have been trying to get my corporate IT to shift focus from Sun Sparc Servers to Sun AMD Operton Servers. I am getting alot of resistance from the Sun Sas because tests that we have done that have been Solaris vs Dell Linux boxes haven't been Apples to Apples. Sun itself won't do direct comparisons between their Sparc and x86 platform. I would love to see an article aimed at Sun Sparc Solaris vs Sun Operton Solaris vs Sun Operton Linux vs Dell Intel Linux that focuses on business type applications (databases, web serving, mathmatical libraries, etc).Anonymous - Thursday, June 23, 2005 - link
I'm planning on building a new workstation PC, primarily for use with solidworks. I'd love to see some wide comparisons using the workstation benches of single and Dual core Athlon and Pentium systems head-to-head, as well as some video card comparos using workstation class and consumer class cards, to find out where the hard earned dollars are best spent. Ram configurations could also be useful (512 vs 1GB vs 2GB), to see where optimal conditions are.sm198 - Thursday, June 23, 2005 - link
#44 & #45 - I would be interested in seeing those benches.Paul - Wednesday, June 22, 2005 - link
How about some 64 bit benchmarks using the Microsoft beta's. Is that not allowed? It seems like we have seen very little (other than your excellent recent server article) that compares 64 bit performance. And what little there is only looks at Linux. Microsoft made some vague claims on a webcast (64 bit way faster), but I'd love to see some Visual studio 64 bit compiler or SQL Server benches, AMD vs. Intel.
illissius - Wednesday, June 22, 2005 - link
Though I probably couldn't recognize ActiveX if I saw it, the comments work perfectly fine for me with Opera, on Linux. All it seems to be using is javascript. (Which I likewise would like to see done without, so I can middle click it to open in a background tab, which is unpossible with javascript.)asli - Wednesday, June 22, 2005 - link
Anand, how about a benchmark of a dual-P-M server compared to a dual Opteron and a dual Xeon server? A P-M provides very interesting alternatives in the SFF/low noise desktop space, but how about blade servers or even regular servers? If it is technically possible to get benchmarks for a dual P-M, wouldn't it also be similar to the upcoming dual-core P-Ms?Alternately, how about a fanless P-M desktop? It would be interesting to know if it would stand up to the rigors of heavy workloads.
Locut0s - Wednesday, June 22, 2005 - link
I second the Quicktime 7 suggestion of Eric Florenzano. Actually what I think Eric means, and what I mean, is that it would be great to see a review that uses an H.264 video clip as a stress test. Although quicktime 7 is the big name I'm not sure if it's the best for benchmarking.General Smirnov - Tuesday, June 21, 2005 - link
I would like to see some comparison of the heat output and noise of these new cards. CPU's have had issues with too much heat and power consumption, about time it hit graphics cards in a big way?Anonymous - Tuesday, June 21, 2005 - link
i agree on the activeX thing if that's what is preventing this from working in firefoxAkshay Dodeja - Tuesday, June 21, 2005 - link
I would deifnately like to see more of MPEG 4 Benchmarks. HD compressions is catching up. Especially in the area of H264 compression.mdme - Tuesday, June 21, 2005 - link
here's an idea, would you by any chance have a beta or something of the unreal 3.0 engine. if it really has multithreading enabled, it would be nice to see how the X2 or PD/EE compares with single core FX's P4/EE's.Bloinkxp - Tuesday, June 21, 2005 - link
How about some MAME benchmarks? These are very CPU bound.Ron - Tuesday, June 21, 2005 - link
Personally, I'd love to see some sort of benchmarks on chess positional analysis software such as chessbase 9 (with deep fritz 8 as the analysis engine.) I don't know how doable this would be or if there is enough demand to make it worthwhile, but I do quite a bit of this and find it to be very cpu-intensive, since it uses a "brute-force" method of anlysis.karioskasra - Tuesday, June 21, 2005 - link
I'd like the "comments" link not be ActiveX activated. I block ActiveX as much as I can to protect vs embedded spyware and if there's no good reason for you to use ActiveX, maybe you could consider letting it go.DigitalArtMonkey - Tuesday, June 21, 2005 - link
I would like to see some reviews of the Apple Cinema 30 inch display running on a PC. I know that it requires a dual dvi connection and not too many cards outside of workstations provide that.Rocket321 - Tuesday, June 21, 2005 - link
I too would like to see some Gentoo compile times like installing firefox or openoffice.Also dvdshrink would be interesting as it is a very common program.
TomcaT - Tuesday, June 21, 2005 - link
Only suggestion would be to see what the single and duel core chips do in multitasking on something other than Windows (Linux/FreeBSD or some such). I'm a bit curious as to how the windows scheduler affects multitasking performance (concidering the rather common ideas that the scheduler is crap). Having "virtual" processors through HT might alleviate some of the inherant issues in the schduler (as seen on various Intel vs AMD benches measuring HT impacts). I don't know if this has been done before, but it'd be an interesting thing to look into.-TomcaT-
MrJim - Tuesday, June 21, 2005 - link
I would like to see a multithreaded flightsim, most i want to see Falcon 4.0 the Allied Force version that was showcased at some intel shows. ThanksAnonymous - Tuesday, June 21, 2005 - link
I would like a BigMac with large fries and a medium coke. Take care.PeteRoy - Tuesday, June 21, 2005 - link
I would love to see anandtech finally making a video review showing all the new stuff in video and showing the benchmarks process in video.Anonymous - Tuesday, June 21, 2005 - link
WOW! Some of you guys are just listing stuff for the heck of it. More than half of it is useless.AT - Tuesday, June 21, 2005 - link
#44: That would be useless. First of all, both of the OSes are pretty old and even though no one has conducted the benchmarks, people aren't looking for that anymore.Secondly, there shouldn't be too much, if any, performance difference between the two OSes.
hondaman - Tuesday, June 21, 2005 - link
Would be nice, just for the record, to see some xp vs 2k benchmarks using games, apps, etc.dvdshrink - Monday, June 20, 2005 - link
Please include dvdshrink as a stand alone test.ravedave - Monday, June 20, 2005 - link
I agree with reference computers. I would belive the average life of a computer for someone that reads Anandtech is 2 or 3 years maybe? Having a 2-3yr old CPU on all the benchies would be sweet. Most people don't upgrade from a 3200+ to a 3400+...It would be cool if you could build a flash widget to do a weighted comparo for processors.
Display all of the categories you have benchies in and what % the CPU is better than some other CPU. Then the user could tweak the sliders to add up to 100%, for what percent of usesage they use. So if someone only did office stuff they could tweak the slider to 100% office benchmarks and see much better CPU A say 10% is than CPU B. Wehere someone else mught tweak 10%office and 90% graphics and CPU A is only 5% better...Some sort of weighted average...
Anonymous - Monday, June 20, 2005 - link
Screw CPUs!Whem is your "G70 is the king of videocards" review coming?
Andrew - Monday, June 20, 2005 - link
I'd like to see some reviews of raid 5 SATA. Plus i'd like to see those reviews as whats supported by Novell Netware. Sata is getting much more inexpensive and now with Sata II, it should start putting a hurting on SCSI or does it even come close?Aquila76 - Monday, June 20, 2005 - link
YAV for Divx6 and h.264 encoding/decoding. Why not throw in WMVHD, too, now that MS released the HW accel. patches for it?McTwist - Monday, June 20, 2005 - link
I'd like to throw in a vote for seeing more engineering benchmarks (e.g. MATLAB). I work in the aerospace engineering field and use MATLAB a lot.Gracias
crtfanboy - Monday, June 20, 2005 - link
I'd love to see DOS emulation benchmarks. Like, try running one of those late 90s 3d DOS games in something like DOSbox, while doing other crap.RDX_fX - Monday, June 20, 2005 - link
-Gentoo compile time benchmark.(default packages for a stage-1 install, scripted total compile time).
-More Reference CPUs included in benchmarks.
Perhaps one representative CPU at each milestone in common/recent CPU evolution; P3 1.0Ghz, P4 3.2Ghz Northwood, P4 Prescott, Pentium-M 1.6, etc(and AMD, of course).
anonymous2 - Monday, June 20, 2005 - link
thursday, nvidia g70 7800gtx launch, just a guess.Consider reviewing Gigabyte's Ram-disk,(pci slot ddr-ram hard drive) I think this could load games very fast and work well with video editing.
Also, the Antec P180 is an interesting new case that has many good features at a reasonable price. thanks
DMF - Monday, June 20, 2005 - link
Imo, we dont see enough quake III based benchmarks anymore, even though some of the ore modern games, such as Jedi Knight Jedi Academy, are still widely played. Why not have a high res benchmark of something in quake 3 to see whether CPU impacts performance or not, after all, we dont buy an FX57 (im assuming) or whatever near $1000 dollar CPU this is in order to run synthetic benchmarks or games at low res, etc... Maybe you could save some people some money by pointing out in games at higher resolutions at what point a faster CPU becomes useless (3000+? 3500+? 1 meg of cache needed?) and its all graphics limited.smn198 - Monday, June 20, 2005 - link
Interesting. The anticipation exhibited by some people at the announcement of a soon to be released article becomes too great if the soon does not equal tomorrow. Maybe an email alert system upon completion of the article would help. Maybe not giving out information on up-and-coming articles too far in advance would help as well.It seems to be a similar situation to the HL2 release – many people anxiously waiting and perceived lack of information or unexpected delays turn into hostility towards Valve.
Mark Little - Monday, June 20, 2005 - link
I would like to see more price/performance numbers and overall summary charts of all benchmarks in one graph. Something like xbitlabs has showing one CPU as the 100% baseline and then showing how much over and under the "next speed bump CPU" is compared to the baseline. Looking at pages and pages of benchmarks can be a hassle. Sometimes a table of percentages (Anandtech has used this in past) and a graph of all benchmarks on the x-axis and percents on the y-axis can be very helpful.trbader - Monday, June 20, 2005 - link
With the dramatic increase of AMD boards mostly using Nvidia chipsets, how about an article on the Nvidia chipset functions -firewall, active armor etc.- how to set up the firewall, what the settings mean, and how they affect real-world use like surfing, online gaming and even P2P. (Most of that set-up stuff is incomprehensible to "normal" folk)Then run benchmarks on the different setups.
ravedave - Monday, June 20, 2005 - link
How about your own MS Office based benchmark?Like loading a 20,000 page word doc.
How about distributed computing? Are there any of them still left that have work units that are always the same?
How about PDF rendering? I know some of the large ones take a rediculusly long time to load.
What about desktop searching? I know its proabably mostly a function of the disk speed but I wonder if CPU makes a difference how fast Copernic or Google desktop search indexes your disk...
Why not write your own C# program that does something as simple as adding 1+1 100million times... is C# optimized for AMD/Intel...
Dave
BBB - Monday, June 20, 2005 - link
One thing I'd like to see in CPU gaming benchmarks is minimum framerates as in GPU benchmarks, rather than just averages.It would be interesting to see how Intel/AMD platforms affect minimum framerates, and whether these results differ from average framerates or not.
Matt C. - Monday, June 20, 2005 - link
I wouldn't mind seeing some performance comparisons among recent CPUs. You mention faster single core vs. slower dual core, I think would be great across both X2 and Pentium D, a "mini-roundup" of a bunch of the latest CPUs would be very useful.kafka - Monday, June 20, 2005 - link
I'd like to pick up on what ViRGE said. I'd love to see some benchmarks of actual power consumption on various systems "as a whole", rather than just looking at the heat generation or performance of the processors. How many watts are consumed under various use scenerios by competing systems and what percentage of the total is attributable to the individual components? For those of us running our systems 24/7, the energy costs add up.drpepper128 - Monday, June 20, 2005 - link
I'd like to see more Photoshop benchmarks as that is what I am interested in. Also Photoshop is interesting as some things in it are single threaded and some are multithreaded.ViRGE - Monday, June 20, 2005 - link
Here's an idea(and it may just be sleepiness talking): create a suite of common, low-impact applications that would represent typical computer usage(say something similar to PC WorldBench), but instead of measuring time as a way to see raw performance, use it as a chance to measure other system attributes. For example, if I'm just surfing the web while listening to music and writing a term paper, what kind of power draw, noise, CPU usage, and heat output be? A lot of tests you do are all about performance at full speed, but when I'm fumbling around with only minor CPU usage, would a dual CPU system be quieter? A faster CPU(lower percentage of CPU time needed to do the task) versus a slower CPU(doesn't need to be tuned for high-clock, high-power usage applications)? I'm sort of envisioning something similar to previous Cool & Quiet / EIST tests, but with a focus on only partial CPU usage, instead of full/idle testing.Anonymous - Monday, June 20, 2005 - link
I'd like to see more video encoding/decoding tests as well with the Nero Digital [H264] codec.Xboxer - Monday, June 20, 2005 - link
hmmm...NDAs expire on Thursday... Gnomedex starts Thursday...any connection? any announcements coming at Gnomedex, technology wise. I have heard though that Microsoft plans to announce some big things at the conference but I thought that was RSS related...anyway...let's see!
(maybe they'll announce that the Xbox will have a PPU now :) )
or AMD has switched to Unified Shaders...or ATI has given up Unified Shaders...:)
Anonymous - Monday, June 20, 2005 - link
Amen to DivX 6 benchmarks and a SATA II article. I'd like to know if they really did fix the connector-breaking issue with the standard revision.Anonymous - Monday, June 20, 2005 - link
Review the Foxconn NFPIK8AA-8EKRS motherboard.wbwither - Monday, June 20, 2005 - link
I know you already run compiling benchmarks with Visual Studio, but I'd like to see a benchmark aimed at web developers -- something in PHP or ASP.NET, using MySQL or MS-SQL. Many small web development shops use their own personal machines for development work, and then once something is done and tested they upload it to the server. So the benchmark could be something simple like an ASP.NET script that makes a bunch of calls to a database, with both IIS or Apache and the database running on the same machine where the web browser is loading the page.... and just see how long it takes to run. Of course, it might end up that the disk or something other than the CPU is the limiting factor, but it'd be interesting to find out. Obviously you're not testing concurrency like you would for a server test, but you're still putting a pseudo-server load on a desktop machine.Creathir - Monday, June 20, 2005 - link
I agree with Demo,Battlefield 2 (demo or the real deal, depending on the scheduled release of your article) benchmarks would be terrific. It would be really great, especially if you throw in some SLI comparisons or even some memory speed comparisons. This game is one of the more demanding games on my system that I have played, and will certainly prove to be a great benchmarking standard for many reviews well into the future. Those are my two cents.
- Creathir
Stephen - Monday, June 20, 2005 - link
Anand -- Call up shuttle and get the SN26P!@We all want to see an SLI SFF!
Ciber - Monday, June 20, 2005 - link
Here's another vote for H.264 encoding/decoding benchmarks. :)Anonymous - Monday, June 20, 2005 - link
I'd like to see some more compiling benchmarks, particularly .net compilers. Most Microsoft shops use C# or VB.net rather than VC++, which is what you're currently using in your development benchmarks.brs - Sunday, June 19, 2005 - link
If I can sneak in "Any requests" for the whole site:Please, please - May we have the SFF roundup you were working on in March and hoped to have up in a few weeks?
Stephen - Sunday, June 19, 2005 - link
Anand -- I would love to see all the game benches maxxed out at 1920x1200 or 1600x1200....If you are going to spend that much on CPU's and SLI rigs -- you are going to play it maxed out not 640x480....
AT - Sunday, June 19, 2005 - link
OTABEL: Go to hell!S.o.N. - Sunday, June 19, 2005 - link
lol at the FX-57 comment...I would like to see maybe ana rticle about Sata II that could quell/confirm all those rumors out there...
Anand Lal Shimpi - Sunday, June 19, 2005 - link
OTABELThursday, I will be able to speak a lot more freely about some of the technology by then due to NDAs presently in effect.
Take care,
Anand
DrB - Sunday, June 19, 2005 - link
I'd love to see more Macintosh articles including perhaps benchmarks and reviews of some of Apple's professional lines of software including the Final Cut Studio series of software.OTABEL - Sunday, June 19, 2005 - link
What the crap?!?!Where THE HELL, is the 360 & PS3 article, wow, this is just RIDICULOUS, this is totally PATHETIC and I just CAN'T believe it.
Again, he gives an update on the article and what now? Another month waiting coming up next?
C'mon Anand, you are better than that, it's also a little hypocritical to just ignore what MANY people are interested about and start talking about the "Yet ANOTHER CPU review"--Please Anand, at least the 360 and PS3 article is something more original, something you don't see everyday, new CPUs come extremely often, I think you know what I mean. Let's do something different, this opportunities don't come very often, so let's get to work man, every day that passes by our (at least MINE) expectations go one step higher, you may as well dissapoint--It would be a first....
Thanks and I hope to see the article.....as soon as possible.
OTABEL
Shopman.co.uk - Sunday, June 19, 2005 - link
How about a look at how amount of RAM affects the Benchmarks. Many people might own this CPU, enthusiasts will load it with RAM (1 - 2GB), System Builders might not (512MB). Would be interesting to see the difference what with memory controllers, hyper this, hyper that...Do large amounts of RAM still offer the amazing performance gains on the same CPU that they were supposed to back in the day or are we all about Video Cards, CPU power and money spending these days?
Lyla - Sunday, June 19, 2005 - link
When I read the title of this piece I thought it was a general "Any requests?" for the whole site, and I would like to make a request that has nothing to do with CPUs:- AMD motherboards, socket 939 PCI-E roundup.
Please... it seems AT reviewd the DFI nF4, was happy enough with it and left it at that. But the DFI is a nightmare for a lot of people, it would be nice (and fair!) to see how the other boards compare for those of us who are not so lucky with DFI.
Anon. - Sunday, June 19, 2005 - link
A few microbenchmarks would be nice. The memory latency and bandwidth benchmarks from LMbench, the STREAM benchmarks, etc.To explore the dual-core performance issues, consider the OpenMP NAS Parallel Benchmarks (http://www.nas.nasa.gov/Software/NPB/) and the HPCC benchmarks (http://icl.cs.utk.edu/hpcc/). Although "batch"-style benchmarks, they show quite some difference between the dual-core systems.
deuce - Sunday, June 19, 2005 - link
can you give me an approximate date to when your article on the ps3 and xbox360 comparison will be posted?illissius - Sunday, June 19, 2005 - link
Well, overclocking, for one ;)Other than that... compiling is mainly all I actually use it for, which you already have. How well it scales on multiple cores/CPUs would be nice to see, though. (In previous reviews, either whatever was compiled didn't support parallel compilation (make -jN), or it just wasn't enabled.)
Demo24 - Sunday, June 19, 2005 - link
I would like to see some BF2 cpu marks. I dunno if that is really fitting into what your wanting but it could put some strain on multitasking. I know it uses a bunch of ram but not so sure on CPU load.Eric Florenzano - Sunday, June 19, 2005 - link
I'd like to see a benchmark of the Quicktime 7 Beta for Windows, seeing as it's so CPU-Intensive, and I have yet to get this working smoothly with 1080p on my A64 3800+. It'd be interesting to know if the FX-57 could handle this better...Oops, I mean, the "next speed-bump CPU review."boban10 - Sunday, June 19, 2005 - link
i would like to see the bench of real media producer (i asked it few times already), and a new Divx 6 codec, but not normal version, but a DivX Helium version: http://labs.divx.com/archives/000055.html(that one is optimized for dual-core cpus)
well thats all, i find more interesting dual-core cpus anyway...