Slashdot Mirror

If they made their processors slower, then they would be "innovating"? What do you want processors to do, really?

Lower power consumption? Lower heat output to the point where I don't need a Rube Goldberg Device?

Intel already has shipped strained silicon products. I have one.

http://www.xbitlabs.com/articles/video/display/gpu -consumption2006.html

NVidia's cards used to be the ones that sucked the most watts and still weren't the best performers. Now it's ATI! Ugh... Fortunately NVidia's got the best Linux drivers. ;)

According to today's X-Bit labs benchmarks, 2GB of RAM doesn't make any difference from 1GB of RAM. It's a total waste.

So we're not even needing more than 1GB in 2006. I'd say 2GB will be needed in another five years, but honestly, why would most desktop users ever need more than 1GB or even 512MB? So we might be at 1GB for a long time.

That post was border-line flame-bait, but maybe the poster was just unaware. ATI has (finally) released a dual-card solution for their x850 line of cards. You can use any x850 card with a x850 Crossfire "Master" card to use it. It's slightly more flexible than Nvidia's solution but this limitation has been lessend with the latest Nvidia drivers. ATI's implementation uses a forked DVI cable that linkes between the 2 cards. It's mode of operation (frame rendering) is configurable bewtween a variety of methods and may also be considered more flexible than the methods Nvidia uses. You decide.

it could be interesting to pair this with
another recent news from Intel, announcing 450mm wafers:

http://www.xbitlabs.com/news/other/display/2005120 1210944.html

45nm transistors on 450mm wafers!

this (roughly) means ( 0.45 m / 0.45 *10^-7 m)^2=10^14 devices on a single wafer!

10^14 is a hundred trillion (not quite the desired 10^15, but getting close)

I haven't seen any articles confirming if Merom will work with the earliest Yonah motherboards, but some articles have confirmed that Merom will be compatible with this chipset:

• Intel Merom to Be Pin-to-Pin Compatible with Yonah
• Fall IDF 2005 - Day 1: Coverage of Everything

From Anandtech's link:

One thing that we found very interesting is that Napa also appears to be the platform of choice for Merom. In fact, Intel's platform for Merom is listed in their literature as either being Napa or a Napa Refresh, not a brand new design. We are hoping this means that Merom will work in Yonah motherboards, also hopefully meaning that Conroe will work in the next-generation Pentium D motherboards.

What does surprise me is that they haven't come up with a better product name to replace it.

>>I really don't see Sony having some wunderkind-middleware that is going to cut this back either.

You mean like a complete API to do all your in game physics for you?

Sony Licensees AGEIA's Physics Technologies.
AGEIA PhysX SDK to Be Used by PlayStation 3 Developers

The beauty of this solution is you can use the same physics API for PC games and possibly even xbox 360 games.

Should really help speed up game development and the ability to port your games to multiple platforms.

Yes, I really do think that. Why? Am I just grabbing numbers out of thin air? Nope.

http://www.xbitlabs.com/articles/video/display/rad eon-x1600_7.html

The X1600 actually outperforms the GeForce 6800. The GT leaves it in the dust though.

My two and a half year-old 2.6Hz P4 with 1GB of ram and a Radeon 9800/128MB runs it just fine with all the bells and whistles. And I'm not sure they even had 6800's back then.

Virtually all games are single-threaded, but a surprising number of them released recently are multithreaded.

Quake 4 and Call of Duty 2 are both multithreaded via patch, thanks most likely to their development on the XBOX 360.

Serious Sam 2 also ships with multithreading support.

That's a whole lotta big-name multithreaded games for winter 2005. I expect many more next year.

http://www.pcper.com/article.php?aid=197&type=expe rt
http://www.anandtech.com/cpuchipsets/showdoc.aspx? i=2668
http://www.hothardware.com/viewarticle.aspx?articl eid=767&cid=1
http://www.xbitlabs.com/articles/cpu/display/athlo n64-fx60.html
http://www.lostcircuits.com/cpu/amd_fx60/
http://www.bit-tech.net/hardware/2006/01/10/amd_at hlon_64_fx-60/1.html

Not until Intel gets the memory controller on the CPU, which isn't until 2007.

Huh? http://www.xbitlabs.com/news/cpu/display/200506152 32538.html

But Intel still charges $2,000 to $3,000 for the things. Currently, AMD Opterons are the best all around chips in terms of price, performance, and power consumption. I can't wait until massive multi-core processors are commonplace, x86 finally dies with disco, and BIOS goes away with Reagonomics.

But the benchmarks show that the PD 820 is equivalent to a Single core A64 3800+ the dual core, being almost twice at fast is at least 90% better than the P-D 820 at multi-threaded apps... and at least 35% better at single threaded applications... well there you go the PR numbering has become askew... a 3800+ is really running on a par with a pentium OCed and liquid cooled to 4.8 ghz. the latest FX processor is really running applications as fast as an overclocked and nitrogen cooled 6.0 ghz pentium. kinda sad. an air cooled AMD chip is out perfmorming the fastest overclocks using nitrogen for the pentium...

If you'd been following benchmarks you would have Known this. Until intel launches a 64-bit processor core they are screwed, they are a generation behind AMD, and AMDs chips will Keep Getting faster for another 2+ years before they need a new generation of core...

now, intel could have skunkworked up a 64-bit chip, with x86 backwards compatability, that runs 32-bit code as fast or faster as it runs 64-bit code... but barring that they're going for hype to fool the unsuspecting masses. by a year or two AMD will be rolling out FX chips that perform equivalent to a 8-10 GHz intel... their 'discount' chips will still be running faster than a 4ghz clocked pentium (reguardless of what their pr numbers are) AMD saw intel's road map, and engineered chips that would keep them in the game.

Intel's roadmap was dreamed up by their marketing department who fucked around with the engineers attempt to make something feasible happen.... i thought that maybe things had gotten turned around at intel, but clearly no. the only thing their engineers have managed to do is build lower and lower power chips with higher performance for the notebook and pda markets. AMD owns the performance and gaming and server markets thanks to intel's 'series of unfortunate events' but intel will retain the performance per watt crown in notebooks, blades, and hand held computing devices... and intel will of course retain the ability to produce high volumes of those chips... no matter how good AMD is, without 10-20 more fabs the size of their largest dresden fab, they simply can't produce the volume of chips needed to 'replace' intel. good news though is that they're going to build maybe 2-3 more fabs, which will put them at the capacity to output ~20% of the chips the market currently buys...

hey, even the inquirer printed that intel would have 6 ghz chips in 2005, and 10 ghz chips by 2010, and clearly amd took it as a serious piece of news, and planned to if not beat them to the punch, deliver products that could compete with what intel was supposed to be offering.

kinda like AMD was the first to deliver 1.0ghz chips in volume 9 months before intel's 1.0ghz pentiums were available in volume, even though intel had announced them as launched before the amd chips.

Intel is still flush with cash, they could always turn things around, but it would take someone like donald trump to say 'You're fired' to all the marketing and managment people who got intel into this mess in the first place. otherwise they're going to keep loosing markets to amd.. and amd will keep getting investors to help them build many many more fabs, so they can replace intel with vendors like dell.

> around 88% efficiency (depending on the model)

If you read some real powersupply reviews (where the majority is real data, not press photos) on X-bit labs such as this one or this one you see that normal PSUs are more like 70-80% efficient in their good range, with only one hitting 90% efficiency. The problem with switcher PSUs is that below a certain power draw their efficiency drops off significantly. For these units it's around 100w. It'll be interesting if X-bit does a review of this unit to see what its efficiency curve looks like.

> around 88% efficiency (depending on the model)

If you read some real powersupply reviews (where the majority is real data, not press photos) on X-bit labs such as this one or this one you see that normal PSUs are more like 70-80% efficient in their good range, with only one hitting 90% efficiency. The problem with switcher PSUs is that below a certain power draw their efficiency drops off significantly. For these units it's around 100w. It'll be interesting if X-bit does a review of this unit to see what its efficiency curve looks like.

> around 88% efficiency (depending on the model)

If you read some real powersupply reviews (where the majority is real data, not press photos) on X-bit labs such as this one or this one you see that normal PSUs are more like 70-80% efficient in their good range, with only one hitting 90% efficiency. The problem with switcher PSUs is that below a certain power draw their efficiency drops off significantly. For these units it's around 100w. It'll be interesting if X-bit does a review of this unit to see what its efficiency curve looks like.

And according to this article (Consider the source carefully), nVidia, ATI, VIA, and SIS are pretty close, just barely past the double digits in percentage. Intel still sells more mainboard chipsets than all others combined. This is for Intel chipsets, and it's because of Intel's own shortcomings that the four main competitors have experienced any growth. I have no doubt that nVidia is strong in AMD chipset sales. However, the number of AMD systems out there, while growing, is still puny compared to Intel systems.

So again, I fail to see how nVidia is even close to becoming a monopoly.

And in response to your comment, I agree--ATI's southbridge sucks. In fact, I'm not too fond of any current ATI product on the market now.

Right. But assuming that the timings on the memory are identical on the two systems, the amount of time to fill a cache line should be the same, from the DDR side of the pictures (tCAS, tRAS, all those good things - all at 200MHz which is 5ns signalling - and at DRAM speeds - also measured in double digit ns). The difference between the two will be the additional time between the CPU (Athlon64 or P4) issuing the command and it being finished.

According to one benchmark, the Pentium4 has a latency of ~64ns and the Athlon64 FX is ~57ns. A difference of about 7ns. Now, add that to the amount of time it takes to transfer all the data. Even if the data transfer were instantaneous, that would be around a 12% increase. Since the transfer actually will take some time (assuming that it takes the same time on both CPUs - it's basically the bandwidth to/from main memory), that cuts that increase down a bit. Suppose the transfer was 50ns itself (it'll actually take longer than that). That's 114 vs. 107. Now we're down to 7% increase of total cache line fill time. What really looks scary is if you translate that into CPU cycles. The Athlon64 running at 2.4GHz spends around 256 clock cycles waiting that long while the P4 at 3.6GHz spends around 410 clock cycles waiting. 410 looks a lot bigger than 256 but it is nearly the same amount of time (114ns vs. 107ns). of course, this is what kills performance... high clock speeds waiting. Pretend the P4 could have an IMC that was just like the Athlon64s. That would save it around 26 clock cycles of waiting. Since the P4 is quite narrow, those clock cycles that are spend doing nothing are what kills it. If an instruction stream that was 100 instructions long, all of which can be retired 1 per clock on each CPU, comes along before having to wait on main memory, the P4 would be 100/410 efficient while the Athlon64 would be 100/256 efficient. That's what hurts the P4.

So, while the IMC helps, I would still maintain that it is not the end-all, be-all of memory performance. Just using the numbers from some web site, we get around 7%. I still maintain that what kills the P4 are just high clock rates compared to main memory and this is really bad because of the nature of program codes. We see that streaming things on the P4 runs fast. This is because there is lots of overlap between computation and waiting on memory. More "normal" codes have lots more jerks and starts because of branching and that type of code will kill a P4 while the Athlon64 will handle it much better.

Why is it that people who are trying to make pathetically weak arguments stick, always resort to personal insults, as if that was going to make what they say any more valid.

Well, in the spirit of quid pro quo (look it up, its latin):

I've been working in network security longer than you've been out of diapers, assuming you are out of diapers.

WPA , like all of the feeble attempts at wireless security, was cracked before it even got widely deployed. For the curious,here's just one of many examples one can find with a simple google search-
www.informit.com/articles/article.asp?p=369221

Like I said before, when you think wireless, think of hoola-hoops. Wireless will be a fad for about a year, then in five years we'll all be sitting around asking "What ever happened to wireless?"

Walmart has cancelled theor wireless plans
http://findarticles.com/p/articles/mi_zd4168/is_20 0307/ai_n9518993
Intel is cancelling intetgrated wireless support plans for desktop chipsets
http://www.xbitlabs.com/news/chipsets/display/2004 0928023130.html
The Starbucks wireless provider went belly up
http://www.theregister.co.uk/2001/10/11/starbucks_ wireless_wonder_goes_titsup/

The wireless wave has crested and is now rolling back into the sea of oblivion.

Like I said before, when you think wireless, think of hoola-hoops. Wireless will be a fad for about a year or two, then in five years we'll all be sitting around asking "What ever happened to wireless?"

Please check the FACTS before making statements like that.

Xbox have been profitable for over a year now.

Okay. According to this page, at full-tilt the Pentium D 820 consumes 130.6W, while this page says the Athlon 64 X2 3800+ consumes 89W. So, how long would the Opteron have to run at full blast to make up the difference in cost of $87? Last month I paid $0.078 per kilowatt-hour. This seems to be reasonably average for the United States. 130.6W - 89W = 41.6W difference between the two. Some back of the Google-calculator math reveals: (US$ 87) / (41.6 W * ((US$ 0.078) / (kW * Hr))) = 3.05871582 years. A not-insignifigant amount of time. If you're in an area where electricity is more expensive like New York or California, the amount of time is even less!

Feel free to correct my math!

I learned from this old Slashdot comment that LCD timings are highly misleading. The '3ms' number means something quite different from what you think it means. In short, see this article, or this forum topic. I've reposted the contents of the latter below.
-------
"Quoted response times by manufacturers are largely meaningless and misleading. .....because it measures the time it takes for full white to black or full black to white pixel transitions. So unless you have your monitor set to maximum brightness & contrast (so that the picture is so bright it burns your eyeballs out) and only use your monitor for flipping blank screens from white to black, and back again, whether the monitor has a 8ms response time or 100ms response time, it doesn't mean an awful lot.

It's the same reason why monitors based on the 20ms Hydis panel outperform the 12ms Samsung panel, the 16ms AU Optronics panel, the 16ms LG/Phillips panel.......

In real world use, the vast majority of monitors (over 95% of them) don't perform anywhere near the quoted response times. That's why you see streaking on the 12ms Samsung panel - its performing at 25-30ms.

Let me try and explain further.

Look at the response times for the so called 'fast' Samsung 172X which is based on a '12ms' panel:-

http://www.xbitlabs.com/images/other/samsung-2/gr2 -2.gif

Since most people have their monitors set to medium brightness (about 80-180 on the grey level scale on the graph) and many applications - particularly games use grey to grey pixel transitions (or one colour to another colour) - the typical response time is somewhere between 25-30ms. Not quite 12ms is it?

Now look at the same response time graph for the Acer AL1721 - a mid level TFT with claimed 16ms response time:-

http://www.xbitlabs.com/images/other/response-6/a2 1-grey.gif

The graph is much flatter, so across brightness and contrast levels, you're going to get consistent response times. At most common user settings, the "slower" 16ms is actually faster than the "quicker" 12ms panel.

Not quite as straightforward as the manufacturers would like you to think. The problem is, by that time, most people have parted with their money. When I was first looking to buy a TFT monitor, I thought that Kustom PCs were a bit mad to stock the Acer monitors in preference to others. However, it's only on further examination that you discover they perform very very well in games - for example, the AL1731M is based on the Hydis panel - and will in fact, outperform the so called 'faster' TFT panels.

From Toms Hardware Guide:-

"For games, the Hydis 20ms panel is still the one to beat. It's not yet perfect, but we know of no other that is faster (based on our tests, of course, and not manufacturers' specifications). Once again, we must insist strongly that the manufacturers' specifications are not to be trusted. "
http://graphics.tomshardware.com/display/20040326/ lcd-08.html

"The response times suppliers associate with their panels vary, anywhere from 16 ms to 25 ms. The only problem is that these figures mean nothing. Or at least, not a lot. An article published in 2001 that can be viewed at Xtremtech explains the situation pretty well, and we have summarized it for you in the section entitled "RT between colors". But this isn't the only problem..."
http://graphics.tomshardware.com/display/20031105

I learned from this old Slashdot comment that LCD timings are highly misleading. The '3ms' number means something quite different from what you think it means. In short, see this article, or this forum topic. I've reposted the contents of the latter below.
-------
"Quoted response times by manufacturers are largely meaningless and misleading. .....because it measures the time it takes for full white to black or full black to white pixel transitions. So unless you have your monitor set to maximum brightness & contrast (so that the picture is so bright it burns your eyeballs out) and only use your monitor for flipping blank screens from white to black, and back again, whether the monitor has a 8ms response time or 100ms response time, it doesn't mean an awful lot.

It's the same reason why monitors based on the 20ms Hydis panel outperform the 12ms Samsung panel, the 16ms AU Optronics panel, the 16ms LG/Phillips panel.......

In real world use, the vast majority of monitors (over 95% of them) don't perform anywhere near the quoted response times. That's why you see streaking on the 12ms Samsung panel - its performing at 25-30ms.

Let me try and explain further.

Look at the response times for the so called 'fast' Samsung 172X which is based on a '12ms' panel:-

http://www.xbitlabs.com/images/other/samsung-2/gr2 -2.gif

Since most people have their monitors set to medium brightness (about 80-180 on the grey level scale on the graph) and many applications - particularly games use grey to grey pixel transitions (or one colour to another colour) - the typical response time is somewhere between 25-30ms. Not quite 12ms is it?

Now look at the same response time graph for the Acer AL1721 - a mid level TFT with claimed 16ms response time:-

http://www.xbitlabs.com/images/other/response-6/a2 1-grey.gif

The graph is much flatter, so across brightness and contrast levels, you're going to get consistent response times. At most common user settings, the "slower" 16ms is actually faster than the "quicker" 12ms panel.

Not quite as straightforward as the manufacturers would like you to think. The problem is, by that time, most people have parted with their money. When I was first looking to buy a TFT monitor, I thought that Kustom PCs were a bit mad to stock the Acer monitors in preference to others. However, it's only on further examination that you discover they perform very very well in games - for example, the AL1731M is based on the Hydis panel - and will in fact, outperform the so called 'faster' TFT panels.

From Toms Hardware Guide:-

"For games, the Hydis 20ms panel is still the one to beat. It's not yet perfect, but we know of no other that is faster (based on our tests, of course, and not manufacturers' specifications). Once again, we must insist strongly that the manufacturers' specifications are not to be trusted. "
http://graphics.tomshardware.com/display/20040326/ lcd-08.html

"The response times suppliers associate with their panels vary, anywhere from 16 ms to 25 ms. The only problem is that these figures mean nothing. Or at least, not a lot. An article published in 2001 that can be viewed at Xtremtech explains the situation pretty well, and we have summarized it for you in the section entitled "RT between colors". But this isn't the only problem..."
http://graphics.tomshardware.com/display/20031105

Really silly artical, they don't know about overclocking Semprons:
Because the latest e-core process runs so cool Sempron 3100 can be overclocked to the speed of a Athlon64 4000+ without needing any special cooling:
http://www.xbitlabs.com/articles/cpu/print/sempron -3100e.html

This is a typical PSU review, that is to say worthless. The problem is to do a good PSU reivew you actually need quite a bit of hardware, most little online sites lack even the most basic testing tools (a good multimeter and a controllable load). They make no mention of how they measured the voltages (software, or voltmeter, and from where, pigtail, ATX connector, somewhere else), they put a system that probably doesn't draw 125W DC at load to test out a 500W PSU, they have no real PSU temperature or efficency information. Typical of a site who's reviewing expertiese consists soley of swaping out parts, running 3D Mark and reporting the difference.

Silent PC Review does half way decent reviews, and over the last year or so XBit Labs has starting doing very good PSU reviews. Beyond that there aren't too many places that consistantly hit the mark.

For a silent PSU (not sure why this is that big of a deal, I have a TruePower 330W and can't hear it over the HDD, but I guess some people will always pay for that last dB quieter), there's of course the Phantom 300, the SilverStone 'NF' series, a 300 and a 400W version, the Fortron Source Zen 300; recently reviewed on XBitLabs and Silent PC Review, with just rock solid voltages across the spectrum. And of course the SeaSonic S12 line while not fanless is known to be extremely quiet and highly efficient

Its already worked, xbox360 launch is going to be like psx2 launch. If you want an xbox360 when it comes out for at launch, you should get one now. You won't be able to find the next month.

I suspect they are just following IBM's lead six months later on this one. Box makers have to go with what their customers want. Its too easy to switch vendors if someone else gives you better price/performance/features.

None of thes programs they tested showed any significant difference, but scientific benchmarks seem to show significant improvement. Much smaller, but still detectable improvement in xvid/divx encoding. The 64-bit version of CINEMA 4D also benefits significantly in most cases (page 11).

None of thes programs they tested showed any significant difference, but scientific benchmarks seem to show significant improvement. Much smaller, but still detectable improvement in xvid/divx encoding. The 64-bit version of CINEMA 4D also benefits significantly in most cases (page 11).

Initial production of Cell microprocessors is expected to begin at IBM's 300mm wafer fabrication facility in East Fishkill, New York, followed by Sony Group's Nagasaki Fab, this year.

Is Intel Good(tm) now?

No.

The new line of chips are LaGrande Compliant. LaGrande is Intel's CPU embedded implementation of the Trusted Computing Group's Trusted Platform Module.

So what does that mean?

All of the new CPUs have ID numbers again. Remember the Pentium 3 ID numbers that created so much outrage and backlash? Whell they are back with a vengance.

The new CPUs will hold crypto keys, and they are specifically designed to keep the keys (and encrypted files) secure against the owner. They are specifically boobytrapped to self destruct if you try to read out your own keys. IBM is currently using a a seperate non-CPU Trusted Computing chip and they explicitly advertize the self destruct aspect in their Man in Black Thinkpad TV commercial.

It can also act as a little spy inside your computer - this is called Remote Attestation - a spy that watches all of the software you run and send a spy report to other people over the internet. You are denied any control over this spy report. The only control you have is to turn this system off completely, and if you turn it off then you get locked out of your own files and it is impossible to run or install Trust-using software. In a five to ten years, under Trusted Network Connect, you can even be denied an internet connection unless you activate the system and send this spy report and you have an approved unmodified operating system and approved unmodified software.

It is basically a DRM enforcer CPU, but far far worse.

-

Noper, I have the same card sitting beside me. Well the box anywho, Linux only supported 1/2 of the card (literally (1 166MHz chip and 1 32mb RAM Module))and Windows XP and 98 support 100% of the card. If you flash the card it would work on a Mac also. Sadly tho that is not the last card from 3Dfx.

The Voodoo 5 6000 is the last production card according to this page. The problem with these voodoos tho, they where too frickin expensive to fight with nvidia's GeForce. Oh well, nice to know there is still someone with a voodoo card.

I own a Canopus 3Dfx Voodoo 1 (6mb RAM) and a 3Dfx Voodoo 5 5500

Intel, HP part ways in Itanium 64-bit processor pact

IBM Drops Itanium Support

Microsoft kills Itanic XP

IBM Rolls-Out Chipset to Support x86-64 Xeon Chips, Plans to Cancel Itanium Servers

Intel, HP part ways in Itanium 64-bit processor pact

IBM Drops Itanium Support

Microsoft kills Itanic XP

IBM Rolls-Out Chipset to Support x86-64 Xeon Chips, Plans to Cancel Itanium Servers

Incorrect. SLI can and does run on some Intel chipsets, at the minimum. This was the original place SLI worked, before ever working on an Nvidia chipset (at least outside Nvidia.) (The article is prior to the actual introduction, but that was one of the boards which worked.) http://www.xbitlabs.com/articles/video/display/gef orce6-sli_3.html

At first glance that is always the conclusuon people draw. But Nvidia and ATI do different methods of anti-aliasing depending on the the level that is chosen. In Nvidia's case they use super-sampling, multi-sampling, and sometimes both. ATI uses multi-sampling only. The result is the same level of AA on each card will produce different visual results.

This article goes into depth about the FSAA issue between ATI and Nvidia. Look at page 12 and beyond for the full poop.

At first glance that is always the conclusuon people draw. But Nvidia and ATI do different methods of anti-aliasing depending on the the level that is chosen. In Nvidia's case they use super-sampling, multi-sampling, and sometimes both. ATI uses multi-sampling only. The result is the same level of AA on each card will produce different visual results.

This article goes into depth about the FSAA issue between ATI and Nvidia. Look at page 12 and beyond for the full poop.

...has hardware H.264 codec support.

And this technology is, in part, targeted at low- to mid-range systems and laptops, meaning it's not going to be part of video chipsets that only cost $599...further meaning that it wouldn't be beyond the realm of comprehension, since Apple is already an ATI customer, for Apple to use something like this in a Mac mini-type product, answering the questions of "how could the Mac mini possibly play back HD?" in the Mac-mini-as-HD-media-center Mac-mini-as-iTunes-HD-Movie-Store-player scenarios.

Off-topic? No, the R520 is mentioned directly in the submission, and one of its primary features is H.264 hardware acceleration. This is huge.

'My point was that Intel has not been behind AMD in memory bandwidth in recent memory. At most times, including right now, they are ahead of AMD in memory bandwidth.' On the other hand, they seem to be behind on the memory benchmarks, so claiming higher memory bandwidth seems suspect: http://www.gdhardware.com/hardware/cpus/amd/athlon 64/fx57/003.htm ' As to your comments that memory manufacturers say DDR2 prices aren't going to drop, I could find nothing like that at all. Most news sources say DDR2 prices will drop below DDR prices in the 2nd half of the year. More specific news says things like I mentioned above.' 'But regardless of the reasons, as DDR2 drops in price below DDR, many Athlon users are going to wish they could use DDR2.' DDR prices will rise above DDR2 as DDR becomes phased out: http://www.theinquirer.net/?article=24059 DDR2 isn't coming down much more, and neither is DDR: http://www.xbitlabs.com/news/memory/display/200506 16222515.html The information is widely available on Asian sites, and gets picked up from time to time by sites like theinquirer and xbitlabs. Since it's from the DRAM manufacturers and OEMs, it's better than Tom's, which is notoriously unreliable. If you want to make it easy on yourself, just read digitimes and xbitlabs every day. But to your point about wishing they could run DDR2, people looking for performance will be buying DDR3, a much better technical solution than DDR2, and AMD offers DDR2 next year when DDR finally is at a disadvantage. 'I dunno about Intel copying AMD's plans. I haven't heard anything of it. ' http://www.xbitlabs.com/news/cpu/display/200506152 32538.html From a practical standpoint, memory bandwidth benchmarks show AMD is ahead, so I think the rest of the discussion is somewhat moot.

'My point was that Intel has not been behind AMD in memory bandwidth in recent memory. At most times, including right now, they are ahead of AMD in memory bandwidth.' On the other hand, they seem to be behind on the memory benchmarks, so claiming higher memory bandwidth seems suspect: http://www.gdhardware.com/hardware/cpus/amd/athlon 64/fx57/003.htm ' As to your comments that memory manufacturers say DDR2 prices aren't going to drop, I could find nothing like that at all. Most news sources say DDR2 prices will drop below DDR prices in the 2nd half of the year. More specific news says things like I mentioned above.' 'But regardless of the reasons, as DDR2 drops in price below DDR, many Athlon users are going to wish they could use DDR2.' DDR prices will rise above DDR2 as DDR becomes phased out: http://www.theinquirer.net/?article=24059 DDR2 isn't coming down much more, and neither is DDR: http://www.xbitlabs.com/news/memory/display/200506 16222515.html The information is widely available on Asian sites, and gets picked up from time to time by sites like theinquirer and xbitlabs. Since it's from the DRAM manufacturers and OEMs, it's better than Tom's, which is notoriously unreliable. If you want to make it easy on yourself, just read digitimes and xbitlabs every day. But to your point about wishing they could run DDR2, people looking for performance will be buying DDR3, a much better technical solution than DDR2, and AMD offers DDR2 next year when DDR finally is at a disadvantage. 'I dunno about Intel copying AMD's plans. I haven't heard anything of it. ' http://www.xbitlabs.com/news/cpu/display/200506152 32538.html From a practical standpoint, memory bandwidth benchmarks show AMD is ahead, so I think the rest of the discussion is somewhat moot.

I mean why is it at something like bignum math or compiling a half clockrate AMD or PentiumM can get equal or better wall-time per operation when compared to a Northwood or Prescott P4?

Until recently it was thought the long pipelines were at fault. But the boys at X-bit labs took a closer look at Intel patents and did some detailed performance measurements.

Turns out that it goes further. The long P4 pipes require "replay buffers" to reissue instructions with unresolved dependencies. These buffers more often than not end up causing further performance losses and power dissipation in case of common patterns of instruction dependency.

"Intel has poured billions of dollars down the tube in all sorts of software and hardware ventures that have led absolutely nowhere (at one point they were supposed to destroy nvidia and ATI. We see where that went)"

Yes, let's look where that actually went. Dig this: Intel currently supplies more than 40% of the graphics chipsets in PCs. By comparison, ATI is 27.6% and Nvidia supplies some 18%. Oops, maybe Intel did win that market after all. E.g., see here: X-Bit Labs.

That's the difference between market reality and fanboy/Cringely talking out of the ass. While the fanboy sees some irrelevant detail, like who's got TEH L33T 3DMARK SCOREZ, the business world is more about other numbers.

Intel is all about making a profit and keeping the profit margins. It's making _great_ money dominating the integrated graphics market. It doesn't need to have TEH L33T 3DMARK SCOREZ, it needs to make money. And it does anyway.

"Intel has poured billions of dollars down the tube"? I don't think so. Those dollars brought it to the position of market leader, starting from zero. Seems to me like anything _but_ poured down the tube.

This technique was used on Kenwood CD-ROM drives, using technology from Zen Research. Instead of heads, they were lasers, but the concept is the same.

Here is a review of the Kenwood True-X 72X.

From the article:
"A conventional laser diode goes through diffraction grating and gets split into seven discrete beams, spaced evenly to illuminate seven tracks."

Of course, you can't split a hard-drive head 7 ways, so I guess this doesn't quite apply. Oh well. Move along. Nothing to see here.

Some time this year, AMD and Intel are releasing dual-core CPUs with hardware support for virtualization. To make sure they sell well, both AMD and Intel are contributing quite a bit of software help to get Xen and VMWare fully working on their chips. Apparently this will help will allow Xen to be able to support Windows natively.

Can you say six MILLION dollars?

Two major transitions for Mac: 68K to PowerPC. Next Mac OS 9 to Mac OS X. Now time for third transition. Transition to Intel-based Macs. Developers Now. Next year for users. "Because we want to make the best computers for our customers." No G5 PowerBook yet. Future products can't be build on IBM of PowerPC. Intel has performance and better performance per watt. Intel delivers much better performance per watt. Starting next year the first Macs with Intel processors. Shipping by next WWDC. Mostly complete by 2007 WWDC. Complete by the end of 2007.

More power consumption data:

http://www.xbitlabs.com/articles/cpu/display/penti umd-820_3.html

Which is slightly different from Tom's findings.

...ATI's forthcoming R520, with hardware H.264 codec support.

Imagine a Mac mini or laptop with that chipset...it will enable HD playback on a lot of hardware that wouldn't otherwise support it.

For those who care

AMD has had a clue since at least 2004... the fact is, low power prossors just are not popular enough ;)

AMD is not in business to 'save the earth' they're in business to make money. There is a market for low power chips, especially in blade servers, and notebooks, So AMD has some low power chips. AMD also wants to make money, off of 'gotta have the fastest gamer cpu' crowd etc.

You can make chips use less power, but they will always have to run 'slower' as a result. So wake up and smell the coffee, faster computers need more power.

These guys should learn how to properly test PSs from X-bit labs http://www.xbitlabs.com./