The GeForce's are good cards, I cannot argue. Personally, buying a card based on what is hoped the gaming market to do, seems like a bad idea. Polygons with more models do look good, but I've heard of some problems with Nvidia's T&L. I wish I were more educated. The problem has to do with something static vs. something dynamic. If I have a chance I'll try to dig up some info.
I've heard others drool over the onboard T&L of the GeForce. What I haven't seen is _one_ game that has a substantial benefit from onboard T&L of the GeForce. Perhaps you could direct us towards some benchmarks?
"Microsoft's pushing and development of Windows 2000 gives us a pretty clear picture of their future roadplans for their OS offerings. And in case you haven't noticed, backwards compatibility with old DOS games is not high on their list of things to support."
Windows 2000 doesn't support DOS. Why is that a pretty clear picture of the future? Has any version of NT supported DOS? I bet $5 the follow up to Win2k doesn't support DOS games either.
Also Windows 2000, their next gen NT, has better support for games (such as Quake3) than the previous versions of NT. That does seem clear, they are going to support backwards compatibility.
Every consumer OS from Microsoft has supported DOS. Maybe they are trying to get away from that in the future, I don't know. How well does dosemu work for running games? Another option would be to dual (triple, quadruple?) boot with one of the Free DOS projects that i've seen around.
"Switching your video card to an AGP one was a great move, because you've moved the graphics into another bus."
AGP and PCI are still on the same bus, I think. not completely sure about this one.
"IF you were running your screen at 1024 by 768 at 32-bit color depth, with a 75 Htz refresh rate, you re already moving more data per second than PCI can handle (according to my calcs, that's >160 MB/Sec - PCI does 132 MB/s)... But video is really the most demanding operation in a desktop computer. "
Your bandwidth numbers seem to be a little off to me. Your video bandwidth calculation is assuming that we are shoving each refresh of the video card across the bus. The 160 MB/s you calculated would be needed between the output of the video card and the frame buffer of the video card.
Also, AGP was discounted as an improvement to 2D graphics quite a while ago. Intel has pushed AGP as a 3D graphics solution because of it's ability to store textures in main memory. But this hasn't even panned out, because local memory on the video card has much higher bandwidth _and_ dedicated bandwidth than using the AGP bus to get to main memory.
One good comparison of PCI and AGP is between the AGP and PCI implementations of Voodoo 3 cards. Basically there is no siginificant differences for the games I've seen benchmarked. In Quake3 and the other standard popular benchmarks (not synthetic, just games) AGP had very little benefit over PCI.
I'm sure that AGP has benefits in certain situations but consider that of the top two 3D consumer chip makers: Nvidia puts 32 Megabytes of ram on board and 3dfx makes PCI cards just as fast as the AGP ones.
Why do you care? Why are you so bitter about it? You don't like it. Don't buy it. Don't read about it. Don't care about it. You wish for an economic downturn so people will have less money? What a crock. Why wait so passively?maybe you could send some mail bombs to these assholes who aren't making the world into your utopia. Get a life.
As many things as there are that I love about Slashdot, I really dislike it when stories just vanish, or something is changed *poof* without any record of it having been different.
Thanks for the information. I didn't mean to imply that set associativity was the only factor, just that it was all that i knew of. Sounds like you are doing some neat stuff.
I think you are overgeneralizing based on a few test cases. One thing to keep in mind is that saying a cache (or any kind of memory) runs at 300 Mhz does not fully (or even come close) to fully describing how fast the cache is in real life. Sdram and FPM memory both can run at 66 Mhz, yet Sdram has higher bandwidth and lower latency.
One factor (for cache) is called set associativity. I'm not an engineer yet, and haven't studied this formally, so beware. Higher set associativity is better. Set associativity indicates that the cache can be used more effectively. You could have a smaller amount of cache, with a higher set associativity that would be as fast as a bigger cache with lower set associativity. This is running at the same speed.
Also, retrieving each byte of data has a certain latency listed in clock cycles (at the frequency it runs at). This latency is as important as the actual MHz rating of the cache.
I don't have the numbers on me, but I found a comparison between the latencies of the K6-3's, Celeron's, and Coppermine's. The Coppermines have a very fast cache, faster than the other two. Much lower latency than the K6-3, even though both look equal at first inspection.
Basically what I'm saying is that it's hard to make generalizations about which case is better, other than to say this one processor is better than that another.
I would argue just the opposite. L2 cache is less important in games than other applications. Do you remember the first Celerons with no L2 cache on die or anywhere else?
They benchmarked very poorly in the typical ZDnet benchmarks for Windows. They also performed pretty well in games such as Quake. Here are some benchmarks from Tom's Hardware:
in Quake2: with L2 cache: 50 frames per second without L2 cache: 45 frames per second pentium 233mmx: 32 frames per second
in Business Winstone: with L2 cache: 25 without L2 cache: 19.3 pentium 233 mmx: 19.5
Granted this is only applicable for Quake2 (and the other quake engines more than likely), it still indicates at least one game doesn't depend very heavily on L2 cache. I think the generalization to be made is that L2 cache boosts integer performance moreso than floating point performance.
Check out the Coppermine vs. Athlon benchmarks you were referring to. The ones I've seen indicate that the on die L2 cache helps the Coppermine catch up to the Athlon in integer performance, but not in floating point.
While I don't really agree with the patent on a fundamental nature, I certainly can understand Amazon's motivation. Think about all the competitor's to Amazon, some have been in the business a long time. Do they care about innovation? What do they care about the us? They don't. They would squash Amazon in a split second if they could, and there are people, probably a good number, thinking of strategies to go after Amazon (and other competitors).
If I were going to fight the biggest kid on the street, I wouldn't be checking my moral code of conduct first. I would be throwing a handful of dirt into his eyes.
We need to pursue changes in the patent system. Not get mad at the kids on the playground for playing tough.
The Groundhog is user number 155439. Is this what new user numbers are at? Seems like we aren't getting any new people logging in here. Occasional I check the user numbers on some of the trolls, Bruce Perens., and Hemos.. Seems like none of the user accounts go above 160,000, but they've been in the 150,000 range for a while now.
Well said Trollmastah. The points serve two purposes: For the readers benefit and to reward the poster. If we hang any more rewards on the poster, Karma whoring will become worse.
I'm glad I took the time to click on your post even though it was slashbot modded down to -1. I noticed up above you were auto modded down to -1, too. I wish moderators would moderate for the readers and not to punish people. Maybe moderators shouldn't be able to see who wrote the posts, just content.
Someone please moderate the parent of this up. It's well thought out and well written, imo. And he makes some good points. What else do you need to be moderated up?
Why would the mode you view comments in have any effect on the number of page loads needed? I load the page once. I can read the comments on it. If I want to read new ones, then I have to reload. What am I overlooking?
I don't know if you are trolling or just ignorant.
1. Celerons are not slow by most people's standards. In fact, for many, many applications and benchmarks, they are virtually identical in performance to a P2/P3 clocked at the same speed. Maybe you are talking about the very first Celerons manufactured that did not have on die L2 cache. Those went away pretty quickly.
2. Celerons are not related to AMD chips at all, except that they execute the same instruction set.
3. If I had my choice of a x86 chip right now, it would be the fastest Athlon I could get my hands on. Granted they aren't that much faster than what Intel is offering.
4. AMD is not slow. 5. AMD is quick. 6. AMD is fast. 7. AMD is good.
Quick, someone email The Register. They need to publish these "facts."
Seriously, if what you say is true, and it seems plausible, we are going to have some fast processors. Of course the downside to this is that people may be unwilling to shell out big bucks to buy a fast processor if the cycle is going move so quickly.
Of course running at a fast Mhz can be the result of better design, not poorer as you imply. Sure, people don't get as excited about a new bus, but that's just the nature of the beast. Faster processor speeds are still important. Other parts of the system may be bottlenecks in some situations, but they are not bottlenecks in all situations. There are good reasons for needing/wanting/producing fast processors.
Why does it suck? It lowers manufacturing costs, which means they can make more money and/or offer consumer's a lower price. I'm sorry I just don't see the problem here. Of course the downside (from manufacturer's perspective) is that just like people overclock their cpus, now they will their ide cards.
Instant Message Wars heating up Making history? PALM (spinoff of 3com) Not exactly nipping to buy... (SGI and Cray) Poll: Monitor Size Seagate Barracuda ATA II @ SR and RAID hack follow-up No BeIA on Intel's WA Sexiest Geek Alive AMD price cuts Trade your games Tim O'Reilly on 1-Click patent
Wow, we finally found one story that was posted on Slashdot also. Thanks for playing. I've been reading Ars Technica for a long time, they certainly aren't a Slashdot rip off.
Try not to look "pathetic" by twisting what the original poster said.
700 Mhz Athlon- $260 750 Mhz Athlon- $360 800 Mhz Athlon- $520 733 Mhz PIII - 530-540 (one is at 478!?!) 800 Mhz PIII - 3 listed at $1000, 1 at $800
If you are going to flame, at least take a second to understand what he actually wrote. AMD's are a hell of a lot cheaper than P3s. The 733 P3 is 170-180 dollars more than an Athlon 750, and 280 dollars more than an Athlon 700. Pretty much what the original poster wrote.
Slashdot Mirror
Thanks for the mirror. It was greatly appreciated.
The GeForce's are good cards, I cannot argue. Personally, buying a card based on what is hoped the gaming market to do, seems like a bad idea. Polygons with more models do look good, but I've heard of some problems with Nvidia's T&L. I wish I were more educated. The problem has to do with something static vs. something dynamic. If I have a chance I'll try to dig up some info.
I've heard others drool over the onboard T&L of the GeForce. What I haven't seen is _one_ game that has a substantial benefit from onboard T&L of the GeForce. Perhaps you could direct us towards some benchmarks?
"Microsoft's pushing and development of Windows 2000 gives us a pretty clear picture of their future roadplans for their OS offerings. And in case you haven't noticed, backwards compatibility with old DOS games is not high on their list of things to support."
Windows 2000 doesn't support DOS. Why is that a pretty clear picture of the future? Has any version of NT supported DOS? I bet $5 the follow up to Win2k doesn't support DOS games either.
Also Windows 2000, their next gen NT, has better support for games (such as Quake3) than the previous versions of NT. That does seem clear, they are going to support backwards compatibility.
Every consumer OS from Microsoft has supported DOS. Maybe they are trying to get away from that in the future, I don't know. How well does dosemu work for running games? Another option would be to dual (triple, quadruple?) boot with one of the Free DOS projects that i've seen around.
"Switching your video card to an AGP one was a great move, because you've moved the graphics into another bus."
AGP and PCI are still on the same bus, I think. not completely sure about this one.
"IF you were running your screen at 1024 by 768 at 32-bit color depth, with a 75 Htz refresh rate, you
re already moving more data per second than PCI can handle (according to my calcs, that's >160 MB/Sec - PCI does 132 MB/s)... But video is really the most demanding operation in a desktop computer. "
Your bandwidth numbers seem to be a little off to me. Your video bandwidth calculation is assuming that we are shoving each refresh of the video card across the bus. The 160 MB/s you calculated would be needed between the output of the video card and the frame buffer of the video card.
Also, AGP was discounted as an improvement to 2D graphics quite a while ago. Intel has pushed AGP as a 3D graphics solution because of it's ability to store textures in main memory. But this hasn't even panned out, because local memory on the video card has much higher bandwidth _and_ dedicated bandwidth than using the AGP bus to get to main memory.
One good comparison of PCI and AGP is between the AGP and PCI implementations of Voodoo 3 cards. Basically there is no siginificant differences for the games I've seen benchmarked. In Quake3 and the other standard popular benchmarks (not synthetic, just games) AGP had very little benefit over PCI.
I'm sure that AGP has benefits in certain situations but consider that of the top two 3D consumer chip makers: Nvidia puts 32 Megabytes of ram on board and 3dfx makes PCI cards just as fast as the AGP ones.
Why do you care? Why are you so bitter about it? You don't like it. Don't buy it. Don't read about it. Don't care about it. You wish for an economic downturn so people will have less money? What a crock. Why wait so passively?maybe you could send some mail bombs to these assholes who aren't making the world into your utopia. Get a life.
As many things as there are that I love about Slashdot, I really dislike it when stories just vanish, or something is changed *poof* without any record of it having been different.
Thanks for the information. I didn't mean to imply that set associativity was the only factor, just that it was all that i knew of. Sounds like you are doing some neat stuff.
I think you are overgeneralizing based on a few test cases. One thing to keep in mind is that saying a cache (or any kind of memory) runs at 300 Mhz does not fully (or even come close) to fully describing how fast the cache is in real life. Sdram and FPM memory both can run at 66 Mhz, yet Sdram has higher bandwidth and lower latency.
One factor (for cache) is called set associativity. I'm not an engineer yet, and haven't studied this formally, so beware. Higher set associativity is better. Set associativity indicates that the cache can be used more effectively. You could have a smaller amount of cache, with a higher set associativity that would be as fast as a bigger cache with lower set associativity. This is running at the same speed.
Also, retrieving each byte of data has a certain latency listed in clock cycles (at the frequency it runs at). This latency is as important as the actual MHz rating of the cache.
I don't have the numbers on me, but I found a comparison between the latencies of the K6-3's, Celeron's, and Coppermine's. The Coppermines have a very fast cache, faster than the other two. Much lower latency than the K6-3, even though both look equal at first inspection.
Basically what I'm saying is that it's hard to make generalizations about which case is better, other than to say this one processor is better than that another.
The links don't work for me now even though they are the right address. They take you to the beginning of the story, just click forward four times.
I would argue just the opposite. L2 cache is less important in games than other applications. Do you remember the first Celerons with no L2 cache on die or anywhere else?
They benchmarked very poorly in the typical ZDnet benchmarks for Windows. They also performed pretty well in games such as Quake. Here are some benchmarks from Tom's Hardware:
Office Performance Windows 98
3D Gaming Performance
Notice the difference between a 300 Mhz Celeron with 128k L2 cache and one without.
in Quake2:
with L2 cache: 50 frames per second
without L2 cache: 45 frames per second
pentium 233mmx: 32 frames per second
in Business Winstone:
with L2 cache: 25
without L2 cache: 19.3
pentium 233 mmx: 19.5
Granted this is only applicable for Quake2 (and the other quake engines more than likely), it still indicates at least one game doesn't depend very heavily on L2 cache. I think the generalization to be made is that L2 cache boosts integer performance moreso than floating point performance.
Check out the Coppermine vs. Athlon benchmarks you were referring to. The ones I've seen indicate that the on die L2 cache helps the Coppermine catch up to the Athlon in integer performance, but not in floating point.
While I don't really agree with the patent on a fundamental nature, I certainly can understand Amazon's motivation. Think about all the competitor's to Amazon, some have been in the business a long time. Do they care about innovation? What do they care about the us? They don't. They would squash Amazon in a split second if they could, and there are people, probably a good number, thinking of strategies to go after Amazon (and other competitors).
If I were going to fight the biggest kid on the street, I wouldn't be checking my moral code of conduct first. I would be throwing a handful of dirt into his eyes.
We need to pursue changes in the patent system. Not get mad at the kids on the playground for playing tough.
The Groundhog is user number 155439. Is this what new user numbers are at? Seems like we aren't getting any new people logging in here. Occasional I check the user numbers on some of the trolls, Bruce Perens., and Hemos.. Seems like none of the user accounts go above 160,000, but they've been in the 150,000 range for a while now.
Well said Trollmastah. The points serve two purposes: For the readers benefit and to reward the poster. If we hang any more rewards on the poster, Karma whoring will become worse.
I'm glad I took the time to click on your post even though it was slashbot modded down to -1. I noticed up above you were auto modded down to -1, too. I wish moderators would moderate for the readers and not to punish people. Maybe moderators shouldn't be able to see who wrote the posts, just content.
Someone please moderate the parent of this up. It's well thought out and well written, imo. And he makes some good points. What else do you need to be moderated up?
Why would the mode you view comments in have any effect on the number of page loads needed? I load the page once. I can read the comments on it. If I want to read new ones, then I have to reload. What am I overlooking?
I'm just curious, what kind of modification have you done to your chip? Did you buy a pre-modded one? Is it stable at 750 with a chace dividor of 1/2?
I don't know if you are trolling or just ignorant.
1. Celerons are not slow by most people's standards. In fact, for many, many applications and benchmarks, they are virtually identical in performance to a P2/P3 clocked at the same speed. Maybe you are talking about the very first Celerons manufactured that did not have on die L2 cache. Those went away pretty quickly.
2. Celerons are not related to AMD chips at all, except that they execute the same instruction set.
3. If I had my choice of a x86 chip right now, it would be the fastest Athlon I could get my hands on. Granted they aren't that much faster than what Intel is offering.
4. AMD is not slow.
5. AMD is quick.
6. AMD is fast.
7. AMD is good.
Quick, someone email The Register. They need to publish these "facts."
Seriously, if what you say is true, and it seems plausible, we are going to have some fast processors. Of course the downside to this is that people may be unwilling to shell out big bucks to buy a fast processor if the cycle is going move so quickly.
Of course running at a fast Mhz can be the result of better design, not poorer as you imply. Sure, people don't get as excited about a new bus, but that's just the nature of the beast. Faster processor speeds are still important. Other parts of the system may be bottlenecks in some situations, but they are not bottlenecks in all situations. There are good reasons for needing/wanting/producing fast processors.
Warranty work on a $20 card?
Tech support? Is that when you read how-to's?
Why does it suck? It lowers manufacturing costs, which means they can make more money and/or offer consumer's a lower price. I'm sorry I just don't see the problem here. Of course the downside (from manufacturer's perspective) is that just like people overclock their cpus, now they will their ide cards.
"Anyone else see the problem with this picture?"
;-)
Nope, faster hard drive only loads Q3 faster, doesn't play any faster.
Let's take a look: Ars Technica
Headlines:
Instant Message Wars heating up
Making history? PALM (spinoff of 3com)
Not exactly nipping to buy... (SGI and Cray)
Poll: Monitor Size
Seagate Barracuda ATA II @ SR and RAID hack follow-up
No BeIA on Intel's WA
Sexiest Geek Alive
AMD price cuts
Trade your games
Tim O'Reilly on 1-Click patent
Wow, we finally found one story that was posted on Slashdot also. Thanks for playing. I've been reading Ars Technica for a long time, they certainly aren't a Slashdot rip off.
Try complaining to your OS maker, they created the problem. It was fixed. AMD makes chips FULLY compatible with Windows 95,98,NT,2000 and I'm sure ME.
Try not to look "pathetic" by twisting what the original poster said.
700 Mhz Athlon- $260
750 Mhz Athlon- $360
800 Mhz Athlon- $520
733 Mhz PIII - 530-540 (one is at 478!?!)
800 Mhz PIII - 3 listed at $1000, 1 at $800
If you are going to flame, at least take a second to understand what he actually wrote. AMD's are a hell of a lot cheaper than P3s. The 733 P3 is 170-180 dollars more than an Athlon 750, and 280 dollars more than an Athlon 700. Pretty much what the original poster wrote.