15,000 is a lot more than 50 times room temp. Assuming the room temperature is 26 degrees, thats 26+275 degrees above absolute zero, around 300 kelvins. I think the writer meant 500 times room temperature. I dont think plasma is 1500 kelvins.
And, the award for the worst math skills on Slashdot goes to...
That article showed that lower latency doesn't mean higher bandwidth (and this is only true if your original latency is low enough, mind you!), but it didn't consider overall performance. Latency has indeed an impact on the performance -- look at Tom's Hardware article on performance improvements when Intel's PAT is enabled. All PAT does is lower latency by 2 cycles.
Heh, you'd be surprised with what you'd get if you let a routing progran route 50 ICs on a two-layer board:). There's still a lot of human intervention needed to get good results.
Use an FPGA, and there's no fabrication time, no back-end development time, and the verification time is significantly reduced.
As an Undergrad Comp Eng, I've designed and implemented (some 74xx ICs + several PALs) a similar, smaller-scale design in a 4-week lab project.
The only impressive thing about his cpu is the fact that he only used the 74xx series and eproms. Impressive, because routing between 50+ ICs is a bitch:).
Neither of those has WiFi, which means they aren't comparable devices.
They were comparable in the other specs that the original poster listed as advantages of the Tungsten device. You are right, H1910 doesn't have WiFi, and it's up to one's personal choice/use to decide whether the extra $200 is worth the difference.
The Zire 71 is is thinner, lighter, and cheaper too.
That, again, is not true. Zire 71 is thicker, heavier, and has the same price as H1910. It also has a slower CPU, but has a 1MP camera built in, which is the only feature bonus over the H1910.
And puh-leaze -- these units DO NOT all have the same screen. PDA screens are not all created equal, by any means. Sony Clies have had better screens than HP or Compaq for years, and the current Palm screens are even better.
Yes, that used to be true. But, now everybody uses the same transflective TFT screens, and they all look fantastic.
My experience with the batter life with H1910 has been nothing but good. I use it all the time, and I only have to recharge it once a week.
The RAM argument is pointless - anybody who has used both devices knows the huge difference here. 64MB on a Palm OS device is about the same as 256 MB of RAM on a PPC device in terms of useful storage.
Altough you have a point here, Zire only packs 16MB or RAM, which would make it comparable to H1910's 64MB of "PPC Ram". The ram issue, though, only comes into play if you're installing applications that take more than the available amount of ram, which is not a common occurence. The RAM is usually used up by large data files, such as pictures, dictionaries, music, etc., and those use the same amount of memory on both OSes.
So, other than the low-res camera, and, potentially, some personal preference/comfort of Palm OS, I don't see any advantage of the low-end Zire over the low-end iPaq.
Yes, but, if they'd put the optimization in to only make the Doom 3 *demo* faster, because they knew it was used as a benchmark, you wouldn't support it any more.
There are a lot of "optimizations" you can use if you know in advance what you'll need to draw. You don't even need a 3d engine, you could just pre-render everything and put an mpeg into the driver. I bet that'd be very fast.
It wouldn't make the actual game run any faster, though.
Try Ut2003, new Wolf demo, Splinter Cell.. Well, any relatively new FPS game, in any resolution over 1024x768, with any kind of image-improving feature (AA, AF).
Do you honestly believe that a million people bought R9700 in the first few months after it came out just to get "extra 20fps", or to win a "pissing contest" (another slashdot favourite)?
(interleaving is essentially RAID for memory: it gets benefits because multiple devices can respond in parallel, rather than in series, so the latency penalty isn't incurred twice).
That's not entirely correct, either. He's talking about "bank interleaving" -- multiple banks on the same module all communicate through a single bus, so there's no benefit to having two banks read at the same time, as only one can return the data.
To achieve "RAID for memory", in performance sense, you need channel interleaving, where data for a single cache line is split across 2 memory channels, and data returns from both at the same time.
Depending on the access patterns from the client, this is not necessarily the best way to interleave memory, though -- altough the latency will be the lowest, you could get a better page hit/miss ratio by interleaving memory accross several cache lines.
First, faster video cards are not designed to make you a better gamer, they are designed to make your gaming experience better. If they are not doing that for you, then you're not playing the games that need the improvement, and you don't need the card. Which, I'm sure, is true for a lot of people out there.
On the other hand, ATI sold over 1 million Radeon 9700s in first few months of it being out, so there are definitely a lot of people out there who do need and want the best card the money can buy.
So, that gets us to your question of whether nvdia cheating really makes a difference. Obviously, it doesn't make a difference to you, because you don't want the buy any of the high-end cards in the first place. It should be obvious in the same way, though, that it does make a big difference to somebody who will buy a high end card.
If 9800 and FX5900 have the same price, and speed is what you're after (and it should be, since you're buying these cards), then you want to buy the faster one. The only way to figure out which one is faster is to check the benchmark results (unless you buy both and try them tyourself). If one of the companies cheated in a benchmark, they have tricked you into thinking that you're buying a faster card, while you're really buying a slower one.
Imagine you're picking between two equally expensive cars, and you want to buy the faster of the two. One claims to do 0-60 in 5s, and the other claims to do it in 3s. You'll go ahead and buy the latter one, only to learn later that they were testing the car going downhill while the other was accelerating on level ground! I think enraged would only begin to describe your reaction to that.
No, my problem is that he didn't have an opinion on the subject at all. The topic, in general, is "is everybody cheating in 3d benchmarks", or in particular, "is nvidia cheating in 3d mark 2003".
His post was "I don't want to spend $400 and still lose against better gamers".
I am honestly interested in why people who do not care about the topic feel compelled to post a message stating that fact.
The developer version is not a pre-release, it's the same version with some extra features that let you debug things, change scenes, etc.
As soon as you move the camera away from it's usual benchmark path, you can see that nVidia hard-coded clipping of the benchmark scenes to make it do less work than it would need to in a real game, where you don't know where the camera will be in advance.
As I mentioned in another post, it's a step in the direction of recording an mpeg of the benchmark and playing it at a high fps rate.
Why do you feel obligated to post the "I don't care about the zillion fps in quake"? Do you post a similar message to every story that you don't care about?
This is a big deal to people who care -- it insults the reviewers who spent hours benchmarking their card, and it insults the users who bought/will buy their card. There are people who care, and people who do want the fastest card for a reason, and they are interested to hear from other people who care, and not the people who don't!
Instead of only looking at the pictures, read the whole article before making decisions on whether it's a driver "fuckup" or an intentional optimization.
The short of it is that nVidia added hard-coded clipping of the scenes for everything that the banchmark doesn't show in its normal run, and which gets exposed as soon as you move the camera away from its regular path.
It's a step in the direction of recording an mpeg on what the benchmark is supposed to show and then playing it back at 200 fps.
Totally different demographics between our two countries. The US has far greater numbers of people living in cities...
That's one of those myths about Canada. In fact, 79% of all Canadians live in cities (http://www.canadainfolink.ca/teach.htm), and over 50% live in just 4 cities. If anything, we have 10 large cities and lots of open space in between up here:).
I'd much rather live in a country where the state of government is to ensure life, liberty, and the pursuit of happiness. Much nicer that way, don't have to worry about insane taxation that way.
That's another one. Altough our taxes are higher on average than the US taxes, it's not by much (depending on the bracket you're in), and it's actually less than several states in the US (Ontario taxes are quite a bit smaller than in California).
If you're starting your own business, then our taxes are much more friendly than US. Lower corporate taxes, lower capital gains taxes, higher R&D credits, and refundable tax credits for smaller companies.
So, yes, they're trying to keep us happy over here, as well:).
True, at the moment ATI is in the lead, but keep in mind that nVidia has now changed over to 13(?) micrometer technology.
Radeon 9600 is also made using 0.13u process. The whole "switch to 0.13" has been overplayed, since neither nVidia nor ATI actually develop the process itself, as they don't produce any of their chips.
The only thing that they have to do is start using new libraries when creating their chips, and altough that's not a simple thing, it's also not something revolutionary -- both companies do it almost once every year.
I wasn't arguing the number of pins added, I was arguing that separating input and output pins wouldn't be as efficient as using twice as many pins that are both input and output.
(And, BTW, a standard memory channel is 64-bits wide, data-wise. 256-bit bandwidth is achieved with 4 of those channels.)
Granted, pin count is higher but I think it would be better suited to the graphics people.
The pin count is one of the biggest cost factors. Doubling the number of pins to separate input and output would be very cost-ineffective -- if you can afford to add that many pins, using twice as many I/O ports instead would be a much better solution, since it would double your peak read bandwidth.
Slashdot Mirror
15,000 is a lot more than 50 times room temp. Assuming the room temperature is 26 degrees, thats 26+275 degrees above absolute zero, around 300 kelvins. I think the writer meant 500 times room temperature. I dont think plasma is 1500 kelvins.
And, the award for the worst math skills on Slashdot goes to...
That article showed that lower latency doesn't mean higher bandwidth (and this is only true if your original latency is low enough, mind you!), but it didn't consider overall performance. Latency has indeed an impact on the performance -- look at Tom's Hardware article on performance improvements when Intel's PAT is enabled. All PAT does is lower latency by 2 cycles.
The bandwidth is 6.4GB/s, not 6.4Gb/s. I.e. it's 6.4 giga-bytes, so the widths of the bus is 8 bytes (or 64 bits).
Heh, you'd be surprised with what you'd get if you let a routing progran route 50 ICs on a two-layer board :). There's still a lot of human intervention needed to get good results.
This guy didn't go much further than what you were dreaming about. In fact, he didn't go much further than what undergrads were doing 6 years ago.
He used 50 of the 74xx IC series chips to build this thing. It's hardly an example of how things change, it's an example of how things were!
It would take a little less than a month to whip up what he did using an HDL and a small FPGA.
Use an FPGA, and there's no fabrication time, no back-end development time, and the verification time is significantly reduced.
:).
As an Undergrad Comp Eng, I've designed and implemented (some 74xx ICs + several PALs) a similar, smaller-scale design in a 4-week lab project.
The only impressive thing about his cpu is the fact that he only used the 74xx series and eproms. Impressive, because routing between 50+ ICs is a bitch
Neither of those has WiFi, which means they aren't comparable devices.
They were comparable in the other specs that the original poster listed as advantages of the Tungsten device. You are right, H1910 doesn't have WiFi, and it's up to one's personal choice/use to decide whether the extra $200 is worth the difference.
The Zire 71 is is thinner, lighter, and cheaper too.
That, again, is not true. Zire 71 is thicker, heavier, and has the same price as H1910. It also has a slower CPU, but has a 1MP camera built in, which is the only feature bonus over the H1910.
And puh-leaze -- these units DO NOT all have the same screen. PDA screens are not all created equal, by any means. Sony Clies have had better screens than HP or Compaq for years, and the current Palm screens are even better.
Yes, that used to be true. But, now everybody uses the same transflective TFT screens, and they all look fantastic.
My experience with the batter life with H1910 has been nothing but good. I use it all the time, and I only have to recharge it once a week.
The RAM argument is pointless - anybody who has used both devices knows the huge difference here. 64MB on a Palm OS device is about the same as 256 MB of RAM on a PPC device in terms of useful storage.
Altough you have a point here, Zire only packs 16MB or RAM, which would make it comparable to H1910's 64MB of "PPC Ram". The ram issue, though, only comes into play if you're installing applications that take more than the available amount of ram, which is not a common occurence. The RAM is usually used up by large data files, such as pictures, dictionaries, music, etc., and those use the same amount of memory on both OSes.
So, other than the low-res camera, and, potentially, some personal preference/comfort of Palm OS, I don't see any advantage of the low-end Zire over the low-end iPaq.
What, like an Intel XScale 400mhz processor in a unit that has a better screen and lower cost than a comparable PocketPC device?
That's simply not true. iPaq H3955 has the same screen, the same processor, and more of usable RAM, for $100 less.
And, H1910 is $200 less for a thinner, lighter PDA with the same amount of RAM, same screen, and a slightly slower CPU.
Try iPaq H1910, then. Smaller than most palms, 200MHz XScale (overclockable to 300MHz), 64MB, a beautiful screen..
What did Jornada have that new iPaqs don't?
Yes, but, if they'd put the optimization in to only make the Doom 3 *demo* faster, because they knew it was used as a benchmark, you wouldn't support it any more.
There are a lot of "optimizations" you can use if you know in advance what you'll need to draw. You don't even need a 3d engine, you could just pre-render everything and put an mpeg into the driver. I bet that'd be very fast.
It wouldn't make the actual game run any faster, though.
Try Ut2003, new Wolf demo, Splinter Cell.. Well, any relatively new FPS game, in any resolution over 1024x768, with any kind of image-improving feature (AA, AF).
Do you honestly believe that a million people bought R9700 in the first few months after it came out just to get "extra 20fps", or to win a "pissing contest" (another slashdot favourite)?
It's just the US dollar that has been going down, and the others seem to be going up compared to US dollar.
Here, try comparing AU$ to CDN$. Neither is growing compared to the other.
Or, here's EURO compared to AU$.
You're kidding, right?
ATI to Host First Public Showing of Valve's Half-Life 2 at Electronic Entertainment Expo
(interleaving is essentially RAID for memory: it gets benefits because multiple devices can respond in parallel, rather than in series, so the latency penalty isn't incurred twice).
That's not entirely correct, either. He's talking about "bank interleaving" -- multiple banks on the same module all communicate through a single bus, so there's no benefit to having two banks read at the same time, as only one can return the data.
To achieve "RAID for memory", in performance sense, you need channel interleaving, where data for a single cache line is split across 2 memory channels, and data returns from both at the same time.
Depending on the access patterns from the client, this is not necessarily the best way to interleave memory, though -- altough the latency will be the lowest, you could get a better page hit/miss ratio by interleaving memory accross several cache lines.
First, faster video cards are not designed to make you a better gamer, they are designed to make your gaming experience better. If they are not doing that for you, then you're not playing the games that need the improvement, and you don't need the card. Which, I'm sure, is true for a lot of people out there.
On the other hand, ATI sold over 1 million Radeon 9700s in first few months of it being out, so there are definitely a lot of people out there who do need and want the best card the money can buy.
So, that gets us to your question of whether nvdia cheating really makes a difference. Obviously, it doesn't make a difference to you, because you don't want the buy any of the high-end cards in the first place. It should be obvious in the same way, though, that it does make a big difference to somebody who will buy a high end card.
If 9800 and FX5900 have the same price, and speed is what you're after (and it should be, since you're buying these cards), then you want to buy the faster one. The only way to figure out which one is faster is to check the benchmark results (unless you buy both and try them tyourself). If one of the companies cheated in a benchmark, they have tricked you into thinking that you're buying a faster card, while you're really buying a slower one.
Imagine you're picking between two equally expensive cars, and you want to buy the faster of the two. One claims to do 0-60 in 5s, and the other claims to do it in 3s. You'll go ahead and buy the latter one, only to learn later that they were testing the car going downhill while the other was accelerating on level ground! I think enraged would only begin to describe your reaction to that.
No, my problem is that he didn't have an opinion on the subject at all. The topic, in general, is "is everybody cheating in 3d benchmarks", or in particular, "is nvidia cheating in 3d mark 2003".
His post was "I don't want to spend $400 and still lose against better gamers".
I am honestly interested in why people who do not care about the topic feel compelled to post a message stating that fact.
Please try reading the article in more detail.
The developer version is not a pre-release, it's the same version with some extra features that let you debug things, change scenes, etc.
As soon as you move the camera away from it's usual benchmark path, you can see that nVidia hard-coded clipping of the benchmark scenes to make it do less work than it would need to in a real game, where you don't know where the camera will be in advance.
As I mentioned in another post, it's a step in the direction of recording an mpeg of the benchmark and playing it at a high fps rate.
Why do you feel obligated to post the "I don't care about the zillion fps in quake"? Do you post a similar message to every story that you don't care about?
This is a big deal to people who care -- it insults the reviewers who spent hours benchmarking their card, and it insults the users who bought/will buy their card. There are people who care, and people who do want the fastest card for a reason, and they are interested to hear from other people who care, and not the people who don't!
Instead of only looking at the pictures, read the whole article before making decisions on whether it's a driver "fuckup" or an intentional optimization.
The short of it is that nVidia added hard-coded clipping of the scenes for everything that the banchmark doesn't show in its normal run, and which gets exposed as soon as you move the camera away from its regular path.
It's a step in the direction of recording an mpeg on what the benchmark is supposed to show and then playing it back at 200 fps.
Totally different demographics between our two countries. The US has far greater numbers of people living in cities...
:).
:).
That's one of those myths about Canada. In fact, 79% of all Canadians live in cities (http://www.canadainfolink.ca/teach.htm), and over 50% live in just 4 cities. If anything, we have 10 large cities and lots of open space in between up here
I'd much rather live in a country where the state of government is to ensure life, liberty, and the pursuit of happiness. Much nicer that way, don't have to worry about insane taxation that way.
That's another one. Altough our taxes are higher on average than the US taxes, it's not by much (depending on the bracket you're in), and it's actually less than several states in the US (Ontario taxes are quite a bit smaller than in California).
If you're starting your own business, then our taxes are much more friendly than US. Lower corporate taxes, lower capital gains taxes, higher R&D credits, and refundable tax credits for smaller companies.
So, yes, they're trying to keep us happy over here, as well
True, at the moment ATI is in the lead, but keep in mind that nVidia has now changed over to 13(?) micrometer technology.
Radeon 9600 is also made using 0.13u process. The whole "switch to 0.13" has been overplayed, since neither nVidia nor ATI actually develop the process itself, as they don't produce any of their chips.
The only thing that they have to do is start using new libraries when creating their chips, and altough that's not a simple thing, it's also not something revolutionary -- both companies do it almost once every year.
I wasn't arguing the number of pins added, I was arguing that separating input and output pins wouldn't be as efficient as using twice as many pins that are both input and output.
(And, BTW, a standard memory channel is 64-bits wide, data-wise. 256-bit bandwidth is achieved with 4 of those channels.)
Granted, pin count is higher but I think it would be better suited to the graphics people.
The pin count is one of the biggest cost factors. Doubling the number of pins to separate input and output would be very cost-ineffective -- if you can afford to add that many pins, using twice as many I/O ports instead would be a much better solution, since it would double your peak read bandwidth.
If you live in Canada, the Space channel will be showing it on April 13:
FRANK HERBERT'S CHILDREN OF DUNE Coming to Space