65 CPUs From 100 MHz to 3066 MHz 359
socram writes " Tom's Hardware posted an interesting article, describing and benchmarking 65 kinds of CPUs from 1994 to 2003. Opinions on what constitutes "adequate computing speed" vary greatly from one user to the next. While one person may be perfectly content with an old Pentium 133 system that stores stamp club membership details in a DOS program in "real-time mode", there is another group at the other end of the scale - video fans who must have the latest and greatest and who will clamor for more and more Gigahertz and gigabytes."
going through your own stash... (Score:5, Interesting)
I've managed to dig up an 8086, 286, 386, 2x 486 66's, 2 486 DX4 100's, P75, P100, 2x P133, Celeron 333, Celeron 400, PIII 450, and an Athalon 1800+....
it's a small list, but shows a good history of computing power in itself.
Zro
Genius to some, Madman to most.
Re:going through your own stash... (Score:2)
Re:going through your own stash... (Score:3, Interesting)
What I found suprising in the article was that an 166MMX with a radeon 9700pro can still run Q3A at a reasonable speed on a reasonable resolution. A new PC is not a right investment, a new videocard might be.
Re:going through your own stash... (Score:2)
8080
8085
8086
8088
80188
80186
80286
i860
i890
80386 (sx/dx)
80486 (sx/dx/dx2/dx4) Nx586
6x86
Pent (60/90/100/133)
Pent Pro
K6
K62
PIII Xeon
Athlon (slot A/socket A)
And those are just the intel chips
BTW, anyone have a i432 they want to get rid of?
Re:going through your own stash... (Score:2)
Oh and before anyone jumps on me, I meant to say Intel and x86 clones.
Re:going through your own stash... (Score:2)
I know, neither are/was the K6/6x86/Nx586/etc. I spotted my bad wording too late, oh well.
Re:going through your own stash... (Score:2, Interesting)
Hey, if you have old PC hardware like this, you might consider contributing to the FreeDOS Project [freedos.org]. Part of our goal is to support older hardware like this. Even though you don't find a lot of 286's in people's homes anymore (at least, not turned on) you still find a fair number of 286's in embedded systems.
Please consider testing FreeDOS on your older hardware. You'd be helping out our little project.
Thanks.
-jh
Re: (Score:2)
Adequate speed (Score:5, Insightful)
Use the correct tool for the job; if a pen and notebook or binder will do, use it. No need to use hours and hours to set up a membership database if your club comprises 20 members and have a meeting every first Thursday of the month...
Re:Adequate speed (Score:2, Interesting)
I remember the old days when a 486 DX2 was not a bad gaming machine...
Re:Adequate speed (Score:2)
Aren't "Granda Jones" and "Old man Willson" the characters in that obscene and heavily repeated troll post?
Re:Adequate speed (Score:4, Interesting)
Those days aren't necessarily over yet! My current Athlon XP 1800+ runs Doom 2 slower than my old 486 did. Thanks to Windows XP's horrendous DOS performance, Doom 2 stutters as badly as it did on my poor Cyrix 386 DX/40 (I think) upgrade chip. After which I moved up to a silky smooth AMD 486 DX4/100.
Wow, I just realized that my last 4 CPUs have been AMD:
486 DX4/100
K6 166
K6-2/350
Athlon XP 1800+ Maybe I should buy an old Duron 1GHz to fill in that little gap I left? :
Re:Adequate speed (Score:3, Insightful)
Re:Adequate speed (Score:3, Insightful)
Caveat: Be sure you know what the job is.
If you just want to keep track of you members, then pen and paper fits the bill nicely. However, if you want to keep track of your members *and* have a warm, fuzzy geek experience while doing so, then go for a nice laptop.
Most people don't *need* a top of the range graphics card any more than they *need* to play top of the range games. That doesn't make getting one a Bad Thing though.
Re:Adequate speed (Score:2)
Yes, and I did RTFA, but they had a bunch of bar graphs there, they didn't say anything about the feel of the machine.
Re:Adequate speed (Score:2, Insightful)
Re:Adequate speed (Score:2)
Re:Adequate speed (Score:2, Insightful)
It's just that when you work with geeks you learn that it can't be said too often...
old theory (Score:5, Funny)
There is an old theory to do with penis size...
For the record, I am running a 286
Re:old theory (Score:3, Funny)
Try this script (stolen from someones Usenet signature): The constants in the script are probably in need of revision, since running it on my puny 350 MHz Linux box (SCSI and over 450 days uptime) rates me at a respectable 22 cm, but I'm not complaining
Conversion to imperial units is left as an exercise to the reader.
moral of the story? (Score:5, Insightful)
Re:moral of the story? (Score:5, Insightful)
Granted. I don't see how any word processor can justifiably require a 1.6Ghz processor and 512Mb of RAM. In fact, I think Office 97 on a Pentium Pro 200 was perfectly usable in it's day and is just as usable now, but that's not the whole story. There's a whole plethora of applications which are now commonplace, which weren't even considered feasible ten (?) years ago. I can remember a piece of DOS software on my old 286 which displayed JPEG images. That was it. It took noticeable time just to decode the file, and then sample it down to 320x200 to display on a normal VGA monitor. Nowadays, we don't even consider the decoding process when viewing JPGs.
There's other similar applications - DivX movies, strong encryption, even MP3 audio - which we now take for granted 'cos we've got so much horsepower to play with that processing overhead is no longer an issue. Now we're getting into the realm of PVRs, digital camcorders, encoding real-time video straight into DivX - applications which appeal to ordinary home users, and which require some *serious* megahertz. The games industry provides a convenient milestone - anyone can tell that Quake III looks better than the original Wolfenstein 3D, but more importantly, they can see that they're fundamentally the same thing. It's a lot harder to compare modern video editing software with that of ten years ago, because ten years ago the only people editing movies on their home PCs were masochistic millionaires.
Rather than focusing on all those wasted MHz driving more and more bloated word-processors, consider some of the things we just *couldn't* do with slower hardware, and wonder what we're going to be taking for granted ten years from now. :)
Re:moral of the story? (Score:3, Insightful)
Now? Hell I can run MP3s in the background when I play games. Running Winamp for hours on end only takes a few seconds of processor time. In under 10 years it has gone from MP3s being something I had to take special steps to play, and do nothing else while I did, to being able to use them for background music no matter what I'm doing.
Things like HD video are one of the Next Big Things(tm) that are going to need serious horsepower. I am mostly content with my 1.6ghz system as it can do everything I want.... Except decode HD quality videos. The Windows Media 9 demos on Microsoft's site stutter during playback, I just don't have the CPU power to decode them.
I have a GAMECUBE (Score:2)
The next step (Score:2)
Re:moral of the story? (Score:2)
Well now I can throw together a P3 800 or something and it will load any productivity app really quickly and do almost anything I could want in a minimum amount of time. Now if I print something, even something really ocmplex, the only thing I'm ever wating on is the printer itself.
Also, plenty of what minimalist zealots call "unnecessary bloat" really isn't. For example even a stripped down version of Linux is "bloated" when compared to DOS. DOS was so simple, you could actually move it almost entirely out of the way if you wanted to. Linux always uses some resources since it always provides basic features like memory management and so on. And then you take modern OSes like Linux with X and something like KDE or Windows XP or soemthing. Require massively more resources than DOS. However it's for a reason. People want the functionality that these things provide. I want a GUI, I want a HAL and HEL, I want multi-tasking and so on. This all takes resources.
Also some things reduce usage of one type by increaing it in a different way. Compression is a good example. Compressed stuff, espcially lossy compression like MPEG reduces disk usage significantly. However, it is done at a cost of increased processor usage. A 128k MP3 file does an acceptable job of encoding a 16-bit 44.1khz wave file and does so at about a 11:1 compression ratio. However, I can playback PCM data on even a 286. My old 486 33 had trouble decoding a 128k MP3 in realtime.
OH, and please let's not get on the whole "only consoles should be used for gaming" shit. There are plenty of reasons to want to play gameson a PC.
100 MegaHertz? You were lucky! (Score:5, Funny)
Prediction: This discussion will end up with someone wittering on about punch cards, paper tape, and front panel access to core memory.
Then someone else will recall the Dilbert cartoon where the engineer boasts "I made a database entirely out of zeros because we had no ones".
Oh, damn, done it myself.
Re:100 MegaHertz? You were lucky! (Score:5, Funny)
Incidently, posting in this thread rendered your mod point useless in it. Spoiled kid, you don't know haw lucky you are.
More Gigahertz != more productive (Score:4, Interesting)
Just because I have an old machine doesn't mean I can't make productive use of it. All right, I can't do gaming, but my Pentium 333 machine suffices for everything else. Just make sure you have enough memory to run everything comfortably without swapping. Heck, I'm even running a webserver [cjb.net] on it.
Re:More Gigahertz != more productive (Score:2, Insightful)
Re: Webserver (Score:2)
--Ghz == more productive (Score:2, Funny)
No, but (Score:2)
Or what about something a little more simple, say MPEG-4 playback? Try and playback broadcast quality MPEG-4 (720x480, 30i couple megabits/second). Not happening on that system. Worse still if you were to try something with HDTV specs.
Now, if the system works for you, great. By all means, use what works for you and don't spend money on more power if you don't need it. HWOEVER recognise that there are plenty of applicaitons (more every day) aside form games that demand more power than a system like that has. The continual increase power is not a pointless thing.
Who read this as... (Score:4, Funny)
Thank you, Captain Obvious! (Score:4, Funny)
Re:Thank you, Captain Obvious! (Score:5, Funny)
Newer. (Click next)
Processors. (Click next)
Are. (Click next)
Faster. (Click next)
Than. (Click next)
Older. (Click next)
Processors. (Click next)
Re:Thank you, Captain Obvious! (Score:2)
Twice!
What's this trying to prove (Score:3, Insightful)
Re:What's this trying to prove (Score:2)
Re:What's this trying to prove (Score:2)
One interesting side-effect of this is the spectrum of device interfaces seen in an office (and causing headaches for the staff). big-plug, PS2, & USB keyboards. Parallel, network, and USB printers. Serial, PS2, and USB mice. Parallel, serial, SCSI, IDE, USB, and Firewire scanners, Zip drives, CD drives, etc. The last six years has really cluttered offices with lots of working (often expensive, too) but plug-incompatible stuff. How's that for efficiency.
Re:What's this trying to prove (Score:2, Insightful)
You've never worked on a big project, where compiling takes between 30 minutes and 1 hour have you? (And yes, this is WITH precompiled headers) Compiling is CPU bound, not I/O bound (say like linking.)
Cheers
--
George Moeckel: So let me get this straight -- I need to upgrade my video card to play UT2003 at the same frame rate I was getting with UT on my old video card?
Poho: Right!
Old machines (Score:2)
Heck, a 486 could do all that!
Influence of memory? (Score:3, Informative)
Wait a minute, surely size isn't the only parameter of the memory that matters? Sure, you have to ensure there's no swapping (if you don't your benchmarks are sure to be totally screwed), but apart from that shouldn't memory bandwidth and latency be good enough to ensure that CPU is the dominant factor? Here [arstechnica.com] is a nice article on this.
Re:Influence of memory? (Score:3, Informative)
Memory bandwidth and latency are determined by the CPU in a vast majority of the platforms tested. For example: take an Atlhon with a 200MHz bus... Fit it with DDR400. No matter what is it going to be able to access all that extra bandwidth, since the CPU bus is saturated. Thereby, the CPU becomes the dominant factor even when taking those factors into account.
I'm sure there are exceptions, so feel free to point them out
Re:Influence of memory? (Score:2)
Which reminds me (Score:5, Insightful)
I've recently put together three boxen for family and friends from my spares pile. We're talking 120-150Mhz PIs, 48-64MB SIMM RAM, 1GB drives, quad speed CD-ROMs, 56K modems and 1MB S3 cards, with Win98SE, Word 97, Outlook Express and not much else.
Now, to me and thee, that spec sucks, but to someone that just wants a box for email, browsing and word processing, it does everything that they need to do, as fast as they need to do it.
Sure, I like being able to buy 3Ghz monsters, but you need to sell a lot of systems to make back the the cost of the R&D for them. And given that we should all be aware by now of the environmental cost of computer systems, I'm going to be keeping "obsolete" hardware in service just as long as I can, and thumb my nose at the marketeers who tell me that there are compelling reasons to upgrade other than the magic smoke [astrian.net] getting out.
I'd like more RAM installed, though. (Score:2)
And they still boot faster than XP! (Score:2)
It was on a PII450, but still, even high end PIII systems take double that amount of time to boot 2K and XP.
Gigahertz (Score:5, Insightful)
There's an old saying, if you sit down at the poker table and don't know who the sucker is, it's you. Any gamer would be better off saving some money on CPU and spending it on graphics card, memory and SCSI disks. The PC architecture is so unbalanced that the only thing a top-end CPU is good for is boasting about.
Re:Gigahertz (Score:2)
If you're buying the absolute, top of the line CPU, then yes, you're a fool. Unless you have to do some serious number crunching where time really is money, it doesn't matter that much. Back off a few revs on the CPU and you'll cut the cost by 60-70% for little speed difference.
But once you're down to that point, then cutting back further doesn't do you any good. It may be a drop of less than $10 to the next speed down. And if $10 makes that much difference in affording the computer, I suggest reprioritizing your spending habits.
If you're a gamer, then, yeah... drop some more money on video or memory. Just realize that for graphics, you may as well go for a GF4 Ti4200 or Radeon 9500 Pro or jump up to a Radeon 9700 Pro - which is a considerable gap. But anything in between just isn't worth the money. For memory, there isn't much that needs more than half a gig of memory (right now - it'll grow), unless you start playing with things like video editing.
For hard drives... SCSI? Why bother? I mean... really. IDE isn't the god awful monstrosity it used to be. Modern drives are fast, big, and cheap. If you want a bit of added speed, you can go for a RAID config, but you're going to spend a lot more money than you'll get back in peformance. SCSI just isn't worth it at all for the desktop PC - even for video editing it's questionable. For servers it's another matter, but we're not talking servers here.
Not so fast, sucker! (Score:2)
See this article on AnandTech: http://www.anandtech.com/video/showdoc.html?i=165
In particular, the charts on this page [anandtech.com] that indicate that the GeForce4 Ti4600 scales up with CPUs all the way to the fastest CPU that was available when the article was written.
Faster graphics cards will only be further limited by the CPU, achieving a smaller percentage of their full potential.
The PC architecture is a little less imbalanced than you think. Spending extra money on a video card your CPU can't feed triangles too fast enough is a complete waste of money, too. Sucker.
Re:Gigahertz (Score:2)
Unless you have a whole lot of multi user access like a file server or something I think IDE is probably the better soltution when you take price in to account.
balanced gaming system (Score:5, Insightful)
Even construing your assertion charitably, I have to disagree. There is a sweet spot for each component in a PC. Let's use NewEgg [newegg.com] for a price check on the Athlon XP.
The Unreal Tournament 2003 numbers [tomshardware.com] are with the current video champ, the Radeon 9700 Pro. Notice that they increase linearly with CPU speed (although not price, unfortunately).You can certainly argue that the $120 premium for the most expensive Athlon XP at NewEgg is not worth 20 FPS (i.e., 46% more expensive, compared to 11% faster). I agree. On the other hand, $120 will not buy you an upgrade from a 120 GB "special edition" IDE drive to a comparable SCSI drive. Even if you did spring for a 10,000 or 15,000 RPM SCSI drive, you would be unlikely to experience faster game play.
The problem with arguing that the PC architecture is unbalanced is that the game writers already know that. They limit texture detail, so that your main memory is barely a factor, let alone your hard drive. I recommend the following for a serious gaming system:
Tweak as desired.Where's Cyrix? (Score:5, Interesting)
Re:Where's Cyrix? (Score:2)
well, i did also except to see cyrixes there, since they would have been the only INTRESTING point there. also the new (cyrix derived?) via c3's&etc would have been much much much much more intresting than 65 intel/amd cpu's.. now the article is just impressive but totally useless.
not to mention how much intresting it could have been with different architechtures too, not just pc. now there was nothing exotic even in there. who wants to read about dated ordinary hardware?
Re:Where's Cyrix? (Score:2)
FWIW, I just put together a linux machine based on the VIA C3M266 motherboard and 1 GHz Via (Cyrix) C3. This machine serves as a firewall, file server, web server, and mail server for a small network. It's on all of the time so I thought that a low power solution would be nice. The machine it was replacing was a dual PPro 150, and it is definitely much faster than that, and runs quieter and cooler.
Since it was replacing an existing machine, the process I went through was to:
1) Build a C3 compatible kernel. (The C3 has no "cmov" instruction, a ppro optimized kernel won't work!)
2) Build the appropriate drivers for the the devices on board. (Of biggest concern, VIA IDE drivers and the VIA-Rhine ethernet driver.)
3) Replace motherboard, hook everything up.
4) Profit. (OK, just kidding.)
Anyways, it ended up being a completely smooth transition. I could not find any info about running Linux on the VIA C3M266 before hand, so I thought I'd share my experience.
Re:Where's Cyrix? (Score:2)
But it wasn't really meant to be, when mmx stuff came out cyrix was the last to join in, and then they were too slow. Sure, the 6x86mx 233 was FASTER than the PII 233, but people weren't buying the pII 233, they were buying at 300mhz.
Now Cyrix has been sold to VIA, and they are using it as a budget, low power cpu. It's just on the wrong socket. I probably would have bought a VIA system at one time, but I can't afford INTEL upgrades, and you can't get a VIA for a athlon MB.
Re:Where's Cyrix? (Score:3, Informative)
Re:Where's Cyrix? (Score:2)
AMD's numbers seem a little more on the money (though they claim they aren't directed at Intel chips) but I still find problematic. The problem is if you look at their chips the trend seems to be up the number by 100 for every 66mhz of clock increase. Well the thing is no processor can exhibit better than linear scaling with mhz, it's just a physical impossability. If you double the mhz of a chip, and leave everything else the same you will at BEST double the performace. Realistically, you won't actually get double for a number of reasons. So AMD's PR system is rather misleading.
Intel comes on TOP (Score:2)
Re:Intel comes on TOP (Score:2)
iirc the amd pr ratings are how it would compare to a tbird cored athlon.
486 good enough for some tasks (Score:5, Insightful)
I showed them Win98 on my laptop. They hated it.
Re:486 good enough for some tasks (Score:2)
Even on todays GHz machines, the hard drive still crawls (but a little faster at 7200RPM). The processors are so damn fast now that the disks are just shameful.
Also left out (Score:5, Interesting)
Also forgot the 486SX (worth forgetting). BTW the celeron came AFTER the PII.
Re:Also left out (Score:2)
Actually... (Score:3, Interesting)
Have you ever run DOS 6.22 on a P133? It's blazing fast. If you must run a DOS app, it would almost be folly to run it on anything more than a P133! Of course, the P133 was not the processor of DOS's heyday; the 80386 was. By the time of the P133, the industry had already begun to migrate to Win95 (or OS/2, or Linux, or you favorite OS - no flames needed here) :-)
in "real-time mode"
Sorry to pick nits on Slashdot, but you probably meant "real mode", though I could be wrong. Real-time and real mode are very different animals.
Re:Actually... (Score:2)
BTW, this machine is still in use - it mainly runs various DOS accounting software in Linux (via the DOS emulator) and there's no problem with its performance.
I am amazed at the speed increases (Score:2)
My guess is its faster memory, hard drive, and buss speeds is what really makes the newer systems so much faster when doing things like ripping mp3's. I find it hard to believe that the actual cpu itself is x100 slower. It may also show why wintel and lintel pc's are taking over the unix workstation market. Traditionally only unix workstations had good i/o but this is now changing.
I assumed these x100 and x80 performance gains would be between something like a 8086 and a PIV. Not from hardware I used back only in 97-2001. You should see how long it takes to load kde3 on it.
Re:I am amazed at the speed increases (Score:2)
Then also there are new ways of processing data. Like SIMD subsystems (SSE, 3dnow, Altivec). For some applicaiton like, say, MPEG-4 encoding it can make a HUGE difference, on the order of a 2-4x speedup on the same processor.
One of my facourite comparisons is OS CPU usage. Back in the 486s days, I had Windows 95. It used a significant and noticable amount of CPU time. I had to quit to DOS to play many games and even to decode a 128k MP3 in stereo. Now, Windows 2000 and XP use such a small amount of CPU time it's insignificant. Less than a processor-minute per day. Now I can tell you 2k and XP sure as hell have MUCH higher overhead than 95. However processors are so much faster that the amount of time they need in overhead as a percentage is just insignificant.
So what...? (Score:3, Interesting)
Here is what I use at home:
I kid you not. =)
It's pathetic, I know.
Now, from the list above, can you guess:
I noticed one thing though: (Score:2)
CNET.com did an article online a couple of years ago talking about the cheapest way to quickly increase performance for your computer. Their conclusion: get more RAM installed first. Easy to understand why--with more RAM installed, any operating system will dramatically lower their need to use hard disk space for virtual memory swapping, which can in some cases increase system performance as much as 100 percent.
My home machine runs a slow-by-2003 standard Celeron A 500 MHz CPU, but because the machine has 320 MB of system RAM installed both Windows 98 and Windows 2000 Professional run reasonably quick because both versions of Windows has very little need to do virtual memory paging on the hard drive.
Re:I noticed one thing though: (Score:2)
On the way to wisdom, you are, young padawan!
(Sorry, couldn't resist!)
Re:I noticed one thing though: (Score:2)
I built my first PC for uni in late 98. It was a K6-2-300 which wasn't top then. Dell were selling 600mhz machines by then, however they were still fitting 32MB RAM to some of them.
I spent 1/5th of the cost of the PC on 256MB RAM. It wiped the floor with any of the 600mhz machines at uni.
That 256 lives on in my now 512MB Athlon 1.4ghz, the only part of the machine other than the case and the TV card that's still there. (Most of said machine has been reused in another PC though)
--
Ditto my work laptop. It's a celeron 400 and did have 64MB and Windows 95. My very first job (without telling management) was to quickly buy the upgrade to 192MB (most it'll take) with £20 of my own cash and put XP on it.
In terms of going home at a sensible time it's more than paid for itself.
Wow this is great! (Score:3, Funny)
older P's and 486's (Score:2, Interesting)
Anyway I am in constant search of these lower speed processors 'cause they are perfect for mp3-players and control jobs! in combination with win98lite it's perfect!
Thanks Tom! (Score:2)
The other end of the curve (Score:2)
Running DOS on a pentium? Blasphemy. That's what a 486 is for.
You forget how good old computers were (Score:5, Insightful)
You're just as guilty of diminishing the usefulness and power of old computers. You forget what those computers did in their day. Did a Pentium 133 typically run real-mode DOS programmes? No, you have to go back a few more years. I have a Pentium 166 running NT4 with IE6 and Office 97, which is far more sophisticated than any DOS programme. It does it very well.
Then there's my network server that also does web and mail for the internet too. It's a P75 and running Debian 3.
Performance of old computers doesn't deteriorate with time. They still run the programmes of a few years ago just as well. They even run some of the programmes of today too.
If you had said "286" instead of "Pentium 133", I might have accepted your comment.
All X86 (Score:2, Interesting)
There were much more interesting (and way faster) cpus coming out around the time of the P5.
I can understand them doing it this way, what with the ease of benchmarking (although they even had problems with that, cpus returning 0) but the fact that I can't recall a single non-X86 article on Tom's hardware might have been more of an influence on them than the practical difficulties of benchmarking.
That it completely misses out a generation of cpus (Pentium 2) is also mildly annoying.
Admittedly though, seeing 3Dmark run on a Pentium 100 was quite fun
Old CPUs need a good home (Score:2)
We've donated most of my older systems; I never throw away a working CPU. Most go to my parents (who still have the old family Kaypro II running on CP/M!). I just turned over to my Dad the oldest machine in my fleet, a 233MHz Pentium MMX box. He's using it for his electronics lab.
We've also donated older computers to poor activists in various social causes.
The lowest-end system at Coyote Gulch is a 400MHz Celeron laptop, which my wife uses for web browsing and e-mail.
Come to think of it, my Sun Ultra 10 is running a 300MHz UltraSparc IIi.
Old boxes can be upgraded and kept going for a very long time. My primary Windows box is currently an 800MHz Pentium III; it began life as 400MHz Pentium II. I keep upgrading the video card and the processors; it still does a very nice job of connecting me to the Microsoft universe. The Radeon 9000 Pro video card lets me run Morrowind and other "high end" games acceptably.
Of course, Linux gets the best machines. I recently acquired a 2.8GHz Pentium 4 system with RDRAM, and it's never been sullied by Windows. I use it for software development and number crunching, so the horsepower is important.
My other Linux box, though, is a constantly-upgraded dual 600MHz Pentium III system. It began life as a uniprocessor IBM workstation at 400MHz... in its current configuration, it is my main machine for long-run simulations. I may upgrade it one more time to dual 1.2GHz Celerons.
"Obsolete" is in the mind of the beholder. Look at the charts on Tom's site, and you'll see that the latest processors are *not* as fast as you might expect from comparing MHz numbers.
Consumer culture convinces us that we need something that we don't. While some of us can use the power of the 3GHz processor, the masses have no need for such a beast. This is one major reason the PC industry has slowed so dramatically -- people who have working systems see no need to upgrade. They're smart in tight times, because, for them, there is no reason to upgrade.
Still, in a society where bigger is better, folls will keep running out the door to buy the latest and greatest, if only to be "bigger" than their neighbors.
Interesting but biased (Score:2)
Also, while the Pentium II is mentioned in Intel's history, it's not in the benchmarks. The Celeron used the same core, but with a differnt L2 cache setup (2nd generation Celerons had 128 KB at full CPU speed, while the P2 had 512 KB at half CPU speed).
They call them killer apps. (Score:2, Insightful)
Many people upgraded to 386 to run win31. Not so many went for the 486. Again to use win95 and play let's say mp3 you needed a Pentium. so many people upgraded to pentium.
(notice that i am talking about upgrades, not all those people that bought NEW computers at the time).
Still P-100 was ok for almost anything, nobody upgraded to P-200, or even P-MMX when they came out. But you can't watch movies on a Pentium ( well, maybe a pentium 233 MMX) and those old soyo's didn't come with USB support (not before the MMX family again). So everybody rushed for P-II's.
At this point , you can do (almost) anything a P-4 3 Ghz does on a simple celeron 333. There is no need to upgrade. Since 1999 we are still waiting for the next killer app. (Games? yeah, sure)
Till them my poor celeron, with 768 MB of really cheap RAM and my faithfull G400 will be just fine.
When to upgrade (Score:2)
So my rationale is that I will upgrade when the performance doubles and I will get a chip that I can get cheap, like the 1700. And I always get a new motherboard because they add so much cool stuff (like USB 2.0 and Firewire and video and sound). I got a new motherboard with 5 channel sound and video onboard for $60. So I go from a 266 to a 1700 for $110 with free shipping at Newegg. Righteous!
Re:When to upgrade (Score:2)
I generally won't bother with a new machine until it is 3x as fast as my old one. Thinking back through my home system history it went
386sx-16, 2M / 40M
486DX-40, 4M (later upgraded to 16M) / 340M
486DX/4-133 (upgraded the CPU)
PII-300 64M / 6.4G - went through the following upgrades:
Upgrade RAM to 768M
Added 20G HD
Added a SCSI RAID controller with 32M of writeback cache and two u160 9.1G drives in RAID0
Upgrade CPU to 900MHz Celeron
OC'ed the CPU just a little.
Now I am considering a new machine, just as the prices on something 3x as fast as the last one are becoming reasonable.
Anything less than 3x as fast just doesn't justify all the hassles of upgrading my system (reinstalling all your apps, etc...) ESP on a work machine. Takes me a lot of work to get my systems set up how I like em.
Not very comprehensive... (Score:4, Insightful)
Faster clocks, NOT more efficient processors. (Score:2)
So, for an 18x increase in clock speed, they get a 26x improvement in performance. That means the number of instructions per clock has gone up by only a factor of two, at least for this application. That's not very impressive, considering how much extra hardware they've thrown at it over the years.
Funny people with too much money (Score:2, Interesting)
Take a statistics course! (Score:5, Insightful)
Tom, how about a scatter plot comparing release date with performance? Or a line plot comparing Intel's top performance with AMD's over the years? Maybe put the theoretical Moore's law curve in there for comparison too. The gentle sloping curve of your performance-sorted bar chart is meaningless. It's a waste of our time and yours.
Another example of Tom being a graph ass is last years printer roundup [tomshardware.com]. He created one graph per printer per group of scales. So we get to compare the hp deskjet's [tomshardware.com] speed at standard resolution with it's maximum motor speed, but we can't compare the speed with that of the canon i850 without flipping back and forth to a different page.
What a waste of good data.
Erik
The CPUs (Score:4, Interesting)
Limited, assymetric instruction set.
Small number of registers. Also assymetric in their use.
Adding new features usually meant extending the instruction set in strange ways, adding even more 'special case' registers.
What I'd like to see would be a good benchmark comparison between other similarly sucessful CPUs.
The 68k and PowerPC series come to mind. Curiously, although the 68k had a much cleaner architecture (at least conceptually) the designers never managed to make it run significantly fast (went up to 60Mhz bus-speed. I think internal speed was 120Mhz).
Sorry not very awake atm (Score:3, Funny)
Re:10 years of boring uniformity (Score:3, Interesting)
Re:Where did they get that P-100? (Score:3, Informative)
Re:Smart companies (Score:2)
Re:Memory (Score:2)
DVD software playback minimum (Score:2)
For decent all software DVD playback, it's probably better to get at least Celeron A 466 MHz CPU at minimum; given that WinDVD 4.0 and PowerDVD XP 4.0 support all the known CPU multimedia extensions (AMD's 3DNow! and 3DNow! Professional, Intel's MMX, SSE/SSE2), a Coppermine-core Celeron 566 MHz (which has both MMX and SSE extensions) is a more appropriate minimum CPU for decently smooth DVD playback.
Coding doesn't require a fast processor. (Score:2)
I cannot understand how anyone who is good at what they do cannot afford the equipment that makes them better at their job. Sounds like a brick mason trying to get a builder to use him for his house but saying he needs to borrow the tools. It just doesn't fit.
Besides if you're really doing 3D rendering, coding, etc... for a living it's a perfect time to take a break while you're compiling/rendering your code/scene.
Re:Coding doesn't require a fast processor. (Score:2)
Unless you fall into coding-by-hacking. Ie:
"I think this does what I need it to..."
<hack><hack><hack>
<compile>
<boom>
"No, not that... let's try this..."
<hack><hack><hack>
<compile>
<boom>
Not that I advocate such a method... I may do it from time to time... but I don't advocate it.
That's what I thought. (Score:2)
Continue to save even after you've purchased your new computer (for the next cycle). I did this in college, my first computer a Pentium 133 was $2483 and I saved for a year and a half to get it. The funny thing is when I began saving the fastest you could get was a Pentium 90 and they were around $3500. As I saved prices drop and technology improves. I know this is common sense but saving for what you want is better than wasting 19.8% on a credit card or MBNA "loan". Plus setup a company for your outside work and write off your expenses. I make about 60k/year in contracts (in addition to my full time job) so I write off everything work related. If you're learning and don't have contracts you can still expense your stuff. If your tax bracket is 12% (fed) it's like getting 12% of what you buy for free. It saves me 30%+ on taxes annually.
Re:stupid people, fast computers (Score:2)
The reason high-end hardware costs so little is precisely because "morons with too much money" buy them. If the tech wasn't mainstream, it wouldn't be nearly as cheap as it is.
Surf pricewatch and build yourself a sweet machine, it's not that hard, and I'm sure you can scrape the budget together somewhere, unless you're in true economic dire straits.
Do we really need that Fast and Furious PC ? (Score:4, Interesting)
One day I said "Do I really need a FAST PC to live?",an later "hey, lets reassemble one of those" and i cramped a 16bit soundcard, 100mbps ethernet and a 1.6gig hd to a pentium 166 mobo with 40megs of ram.
Partitioned and Installed(tm) linux and windows95b, tunned the OSs and played mp3s very fine, played Age of empires, browsed internet pretty well and eventually served as a proxy for my house lan.
The PC finally ended as a car mp3 player inside a small cardboard box in my trunk.
The question is: Do we really need Powerfull but Expensive Computers to do what we are used to do on them?
as a side note, check overcaffeinated.net webcomic. Its yet another pretty fun geek comic.
Pedro Meza - Mafufo.com