iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Monthly Hardware Digest
November'2000

December 6, 2000



Processors.

This time let's begin with "light artillery" and come up to "heavy" step by step. It'd be good to begin with Crusoe that was very active this month. The beginning was not so pleasant for Transmeta, because another two notebook manufacturers cancelled production of computers based on it's processors, at least in the near future. So IBM and Compaq joined Toshiba that did the same earlier. The companies say that the use of Crusoe doesn't give any important advantages comparing to old mobile Pentium III, so using of already checked processors is understandable.

It's interesting that all these announces made right before Transmeta's quoting brought almost no harm to it, the company even raised the quote prices from $16 to $18, and they started on the markets even from $21. At least Crusoe based notebooks are still being manufactured by other companies, i.e. Casio announced its new CASIOPEIA FIVA to be producted in '2001.

And maybe new software component of Crusoe will change the attitude of brands to it. 4.2 version is said by Transmeta to increase energy savings by 20% comparing to 4.1. And changing to 0.13 micron TSMC may affect some perspectives as well.

The end of the month was bad for the company - NEC was forced to return some Crusoe based notebooks due to errors in processor work. Not so much, just about 300, but such things are not good for Transmeta that must gain only positive remarks now. As a result quote price fell down to the level even below the starting.

Competitor companies were active as well. For example VIA has good chances to be a winner. It's 600-1000 MHz Ezra that is to be released in the second half of '2001 may have higher performance comparing to Crusoe, and VIA has more resources to push its product on markets.

This processor may even find some use in desktop PCs, if it'll turn out to be better than Cyrix III. By the way VIA has expanded this line in November, adding 650 and 667 MHz versions. There's nothing to comment here, it's known that this prosessor can be used only in beginner-level office machines and in playstations. OEM may be pleased by this fact, but it's market perspectives are more than humble.

But back to mobile processors, now because of another company. On the last Comdex ALi demonstrated mobile chipset for Athlon, and therefore we can be sure that this winter will present us a new Athlon/Duron line. We'll see what positions AMD and Intel will take, and it's early to speak about some results now. It should be noted that copper can give better performance here than in desktop PCs field, and PowerNow! technology looks better than SpeedStep. And this is not surprising because processors using it will be released half a year later.

In common AMD's business is kind of specificial. The company "lost its head" due to much recent success and gave many promises that it's taking back now. I.e. Palomino that was to be released this winter, may only be announced in winter and shippings will begin only in the end of the first quarter. And its "lesser" brother Morgan will only be announced by this time.

And Mustang is cancelled at all. AMD has preferred to divide its niche between Palomino (AMD-760MP + 2 Palomino = very low priced dual-processor system) and 0.13 micron 64-bit ClawHammer whose performance it thrice better than that of 1 GHz Athlon coming out in the end of '2001.

Anyway company looks ahead with optimism, hoping to take up to 30% of x86 processor market by '2003 when Fab30 will be working full power and will change to 0.13 micron technology. We'll see, because Intel will have some of such factories by that time as well, it has officially announced in November that 0.13 micron technology has been finished and Fab20 has began converting for it in Oregon.

Along with changing to 0.13 micron technology Intel will also begin changing to 300mm plates and by '2002 the company will have two such factories already. And this is a serious factor in "price wars", because chip cost price will be lowered greatly comparing to 200 mm plates used by AMD. And 0.13 micron will be serious in this situation as well. I wonder what cost price 0.13 micron Pentium 4 will have, because Intel already has it. And I also wonder what 0.13 micron Celeron will look like by '2002.

And now it can't be changed even to 100 MHz FSB. Celeron 733 and 766 MHz were at last announced in Europe in the middle of November, though selling for some month already in Japan. But this hasn't changed anything, AMD remained leader in prices and clock rates. And nothing is going to change in the near future, because 900 MHz Celeron is expected only in the 3rd quarter of '2001, but 900 MHz Duron will be released much earlier.

But let's leave Celeron aside and talk about main event of this month. Pentium 4 was at last released on 20th of November. However the majority of reviews are rather unpleasant, but this is not a surprise: it has already been known before the release that we have to wait for SSE2 optimized applications and 0.13 micron Pentium 4. So nowadays there are no urgent reasons to switch platforms, and note that the same system based on 1.2 GHz Athlon costs $1000 less.

Intel is aware of situation and is trying to change it for better by lowering prices, planning to release 1.3 GHz P-4 to the 1st quarter of '2001 (and still not planning to produce it in the next quarter). It's a pity but the price of Pentium 4 chip is only one factor among others like high computer case price, motherboard and memory price, and Intel can do nothing about it.

Speaking about brands I can say they're like they must be: slow and sluggish. Itanium has already began to be shipped to the "chosen" among corporate clients - 20 systems are shipped by now, hundreds will be shipped in the 1st quarter and so on. So Itanium is completely released now, but, and it has been spoken about, only as a pilot platform for IA64 architecture.

Alpha Processor Inc. showed some activity as well. It has changed its name to API NetWorks Inc. and has announced new 833 MHz Alpha 21264 with up to 8 MBytes of DRR SDRAM 2nd level cache support. But this is all for future, because shippings will begin only in January, and therefore all other brands like Alpha, Itanium ClawHammer will have to wait for the next year.

Motherboards.

Plenty of news are to be spoken about in this part due to Comdex and AMD-760 and i815EP announces. But as they are all alike, I give you a brief summary here.

In this period when AMD-760 has been already announced, but there are still no motherboards for it, manufacturers decided to release some new variants of already existing products, like KT133A instead of KT133 and 686B instead of 686A. As a result we get 266 MHz FSB Athlon and ATA/100 HDDs supported. They lacks only DDR SDRAM support but it will appear by the time of AMD-760 boards release.

And mass releases will begin only next year, though many announces are made already nowadays. But having in mind that the chipset itself is late due to some errors, and that AMD will primarily try to please OEM demand, we can assume that motherboards based on it will appear for sale by retail only in January.

And this is all for better, not to step after Micron that has cancelled motherboard shippings until the middle of February, due to some problems with Gigabyte boards, where an additional hardware filter has had to be installed to reduce electrostatic noise.

There's not much to say about other chipsets: SiS730S is to appear only by the end of the year due to problems with AGP 4X. All news about KT266 concern AMI's announce about supporting it in it's new BIOS. It's even worse with multiprocessor Athlons - AMD-760MP motherboards will appaer only in the middle of '2001. But its hard to argue here, because it's an important thing for Athlon to reach server & workstation market, so all mistakes are to be corrected beforehand.

By the way, there are more than enough announced boards on this chipset: DFI, FIC, Epox, MSI, Gigabyte, Lucky Star, etc. But still we can only hope that even some of these products will be released by the end of the year.

Perhaps, boards based on ALi MAGiK1 chipset with DDR SDRAM support will be out earlier. However, manufacturers complain about its unstability, and maybe therefore it is held until January. This chipset isn't popular now: only Epox, Iwill, SOYO announced some models based on it.

VIA's business is far from good: it has passed the last round, having placed its bet on Intel's (and with Cyrix III release it's own) platform. Earlier all resources were thrown for PM133 due to KT133's delay, than the same happened with DDR SDRAM chipsets. MSI has become the only to present a prototype based on KT266 on Custom PC World'2000 (though it was not working).

The situation is opposite with Apollo Pro266 chipset, because many manufacturers (like QDI, Epox, FIC, DFI, Lucky Star, Chaintech, MSI, Gigabyte) has already announced models based on it. We seem to get a new "hit" here or all of them are mistaken about it. Though we'll still have to wait for it. And for the most unpatient Aladdin Pro5 chipsets are being announced.

When there's some time until coming of DDR SDRAM, manufacturers try to fill it with business around newly announced Intel's chipset. Everyone had been waiting for 400BX's successor, and when it has appeared at last, every brand like Abit and Chaintech, Asus and MSI has included it in the plans. Announce followed an announce already from the middle of November. Though every model turned out to be a "classical" middle level variant that will hardly change in the next year.

Having in mind that i820 is leaving already in the 1st quarter, Intel has cancelled its Almador to support i815EP with Tualatin support, and though Brookdale belongs to the same class, it's Pentium 4 support makes it stand aside. It's pity that even in the 3rd quarter Brookdale will lack ICH3 - South Bridge with USB 2.0 support, because it's held until the 1st half of '2002. So Intel's roadmap missed some "points" losing chipsets like Tehama-E (i850 with ICH3). And i850 is going forward at a normal speed: all boards have been announced, manufactured, sales have begun, just no complaints at all. Though ASUS is changing something already now for its new P4T 1.06.

There are only two news left. VIA is preparing defenses against NVIDIA that has came to chipset market, having bought IC Ensemble, an audio-DSP american manufacturer. Companies already have the experience of collaborate work - last year VIA has licensed some technologies of IC Ensemble to be used in its chipsets. But the result is not impressive comparing to NVIDIA's promises, but we'll see what happens next and how much MCP-1 will cost.

The last news is about practical changing to a new server bus. IBM has released first PCI-X bridge that remains compatible with usual PCI and raises its parameters up to 133 MHz and 64-bit. As a result we get maximum capability of 1,064 GBytes/sec instead of 133 MBytes/sec, and it means that Gigabit Ethernet, Ultra320 SCSI RAID and etc. are already at hand. However this chips will be manufactured only in March.

Memory

It was a good month for Rambus. In the beginning of it, company has announced that SDRAM is being licensed by one of the brands, Samsung. It's a good money support for the company, but no more, and all other companies will still remain at their positions.

Anyway RDRAM continues to develop. Samsung has presented 3rd generation chips: 0.17 micron, 128, 144, 288 MBit PC800, and PC1000 will be out soon. The company hopes to increase the amount of RDRAM chips up to 20% by the next year, and they say that RDRAM market will extend by more than 10 times by '2002 comparing to '2000 and will reach 9.2 milliard dollars (the whole DRAM market will be 43 milliards by that time).

Micron isn't going to plan 3rd generation 128/144 MBit chips, instead it ships 256/288 MBit chips of the 2nd generation and they will be the company's mainstream for the next two years. 128/144 MBit chips will be obsolete by '2002, and the company has no opinion about 576 MBit chips.

Now back to Rambus. In some days after Samsung's announcement, another memory brand Elpida Memory (NEC's subdepts united with Hitachi's) has announced licensing SDRAM of Rambus. As a result all japanese brands have their licenses now. In South Korea one has it and one is in the court. One of the Europe's first five - Infineon - is in the court too. As well as one of USA first five - Micron.

The situation can be changed by Taiwan manufacturers that united have a large part of the market, and they are not going to do it now in spite of new Rambus office in Taiwan. Winbond, Promos, Powerchip have denied manufacturing these memory chips. They say that demand is too little now and besides there's much pleasant DDR SDRAM.

By the way, DDR SDRAM is supported by another local brands Nanya. In the beginning of '2001 the company begins manufacturing 128 and 256 MBit DDR SDRAM chips as one of the main products. But they unknown to be in open sales, having in mind that Nanya is IBM's OEM partner.

But in spite of Nanya there would be no deficit on the market. Viking, one of the largest memory manufacturers in the world, has began in November the shipping of 128 and 256 MBit DDR SDRAM samples. Such modules were approved by Samsung for usage in systems based on AMD-760. And Micron promises in the second quarter of the next year to begin shippings of 333 MHz PC2600 DDR SDRAM modules in industrial sizes, having completed testing and certification. In addition, in the middle of November Micron has begun to ship 0.15 micron 256 MBit DDR SDRAM units, that gives us DIMM modules up to 1 GBytes. And no more is needed by today's chipsets, taking into account, that the maximum size, supported by them, makes up 4 GBytes and maximum amount of DIMM slots supported is 4.

When vast memory sizes will be necessary (for example, for servers based on ServerWorks chipsets), maybe 4 GBit (!) DDR SDRAM chips (presented by Samsung on ISSCC) will be ready. They are made with 0.10 micron technology and, as the company asserts, they are to be the first in the world to have 4 billion of transistors.

But that's just a science fiction today. Samsung, undoubtedly being the leader in 64 MBit SDRAM chip production, does not mean anything in 128 MBit part of the market - 23% against 43% of Micron. That is easily explained, taking into account, that Samsung still uses 0.17 micron technology, whereas Micron has already passed on 0.14. So here's the result. Moreover, here are Taiwan and Japan - Elpida Memory becomes the first DRAM manufacturer with 300mm plate factory with 0.13 micron technology "installed". Samsung will begin to build such plants only at the end of the next year, and Hyundai has very serious financial problems.

The only pleasant for them are memory prices, which continued falling in the beginning of the month (it was less than $4 for 64 MBit chip), and in the second half of it have unexpectedly made a powerful jerk-up and a week was enough for market prices to rise over the contract prices (for the first time for the last 4 months). So objective factors show that memory prices are rising now.

Video

Everybody has almost forgotten about GF2 Ultra: who had wanted it - has already manufactured it. Right at the beginning of one month late GLADIAC from Elsa has risen and that was all. And the main activity in this month was about GF2 Pro. Some have manufactured cards based on it (like Sparkle, Asus, Garnet or SUMA) and some have refused (like ELSA or Creative). But they were wrong, for Pro's price/performance ratio looks a bit better, especially, if it's tweaked. Whereas GF2 Ultra can't be tweaked any further - there are problems even with regular frequencies.

A new GF2 MX looks nice, because manufacturers decided to use 183 MHz 5.5ns memory instead of usual 6ns. And taking into account that the memory bus is the most problematic for MX, these changes can do only good.

In general, MX seriously grows. This chip has already appeared in notebooks - the company has presented it's GF2 Go on Comdex - the first mobile solution based on GF2 MX. Theoretically, it looks nice, maybe even too nice. We admit TwinView to be helpful notebook users, HDTV support DVD decoding are as well. But whether users are ready to pay for hardware T&L is still unknown. At least after nVidia's announcement of the design won from Toshiba, there were no further victories, whereas there were many such announcements from ATI concerning old Rage Mobility. But there are some other priorities, because T&L is not primary thing, but Mobility's low energy consumption and heat release can please notebook manufacturers.

And now back to MX again. Namely, speaking about AGP-V7100 DELUXE COMBO from Asus that has been released in November. No need to talk about card itself, as everything is already told and nVidia now has the answer to Marvel and All-in-Wonder. It's just to remind that video editing solution market and the process of PC transformation into the digital VCR continues to speed up.

Let's talk about November's events. Gainward has presented VideoXpert: video in/out, stereo audio in/out, 8 MBytes buffer to support realtime editing, special effects overlay with DVD quality subsequent recording to the storage device (hard disk, optical disk). Innovision has announced DV-2000 MediaCombo - a usual FireWire controller but with ATA/100 RAID onboard. Hauppauge Digital has released WinTV-PVR - an understandable title: TV-tuner/digital VCR/FM-tuner. But for that $250 they ask it's easier to purchase V7100 Deluxe Combo.

Even ATI have announced its TV WONDER, but it bears no relations to videocards, it's an external USB-tuner with a heap of different tools (well, only if it works in USA). Equally for the same price, VoodooTV 200 looks more attractive having the same (and a heap of other) functions.

As if to video, ATI has recouped Radeon VE, that has been announced shortly before Comdex. It was expected to "kill" GF2 MX, but it became a killer of G450. It's interesting, but no so pleasant. Everything sounds nice - ATI's dual monitor support (HydraVision technology licensed for Appian), 32 MBytes of DDR SDRAM, $129 price (and $30 as a compensation - $99 in total). But a cut out hardware T&L and only one pixel pipeline left instantly send this card from computer games into office market, where it can compete only with G450, keeping GF2 MX the leader of the game market. So ATI has made a quite good OEM variant, whereas it has Radeon SDR to compete with MX.

Radeon itself in the meantime has been improved - 0.18 micron technology allows to make a chip with 240 MHz clock rate! And as DDR SDRAM with such clock rate is already present on the market, it turns out that only some marketing things spoil the making of an answer to Ultra. But it obviously will be released in June, because R200 is planned for that time. It will already be made with 0.15 micron technology and will have 300 MHz frequency. And taking in account 4 pipelines, it'll show us the performance of 1.2 Gpixel. It's good that T&L will remain in RV200, though the amount of the pipelines would be truncated twice.

But, observing a status of the game market, it's possible to limit to GF2 or Radeon and just wait for the next Christmas. Besides NV25's price will surely fall, and it's said to have fantastic performance (100 GFLOPS, up to 7 times faster than GF2 Ultra), and in November R300 will appear on the market, having a 350 MHz kernel and thrice performance (as asserted by ATI) than R200!

It's a pity Matrox is still inactive. It has skipped G600, and we shall hope in December it will announce G800, but by some preliminary information it will be able to compete only with G256 or Radeon. However, Matrox again frightens us with some "magic" technologies. It may do some some good - G400, in it's time, has been successive due to DualHead and well advertised EMBM.

We shall not forget about 3dfx as well. However, it has nothing to brag with in November. And the release of Voodoo4 4500 for Mac is not a reason. And troubles follow. A long expected decision about Voodoo5 6000 cancellation: compatibility problems (AGP-to-PCI card bridge and simultaneously 4 VSA-100 chips) and, perhaps, the most important problem - high seller's price of this monster with an external power supply unit that makes this card have little market. There was a question: whether it's necessary to spend money, and 3dfx itself has answered it, having sold really interesting (not for the mass game market) technology to Quantum3D.

The facts are: closing of production, that is a correct step back to former profitable model of business, but still unpleasant; the fact that Voodoo3/5 AGP doesn't work with Pentium 4 boards - AGP slot is made in account on 1.5V and the cards are made in account on 3.3V slot. But the most interesting thing is that Voodoo4, made in the universal form-factor, doesn't have such problems. And also that 3dfx has optimized its drivers for SSE2. What for? For using Voodoo5 PCI together with Pentium 4?

However, future's near. We hope the company will live up to it, despite the operational expenditures and lowering capitalization. Eventually Rampage chip samples (VSA-200) are already in the company's pocket, and Sage probably too, so all Spectre package is ready for production. It'd be nice to produce it now to live up to Gigapixel technology (Mosaic architecture), it'd be a real step to "revitalization".

Now some hi-life talks for the end. 3Dlabs seems to have finally lost interest to the market of consumer videocards, has been divided in two parts - one will continue working with professional graphics, and the second will work on integrated solutions: products for auxiliary graphics output, on the plane instrument boards, for example, and so on. The company has decided to become a leader in this market: to generate a special API consisting of OpenGL subsets, etc.

The last news - at the end of one month a draft specification of AGP 8X (version 0.9) has been released. Just to explain it's essence: DDR AGP 4X. That is, we take the same 66 MHz 32-bit bus with 1.056 Gbyte/sec bandwidth, but at the expense of both slopes communication, we double it up to 2.1 GBytes/sec. Not so pleasing, but however, as it'd be the first serious change in AGP specification, we won't see first AGP 8X products until '2008, so we shall get used to AGP 4X for now.

Storage devices

Contrary to usual, the month was interesting enough. There was Comdex and it means many novelties. And of course the heap of different external CD-RW drives. New nice-looking Iomega Predator showed hardly out-of-date 8/4/32 speed. However, this is FireWire instead of ATAPI or SCSI, and USB variant speed is even less - 4/4/6.

The company tries to remain in the common limits. FireWire and USB are generally accepted to not using their peak bandwidths and to write/rewrite speeds underestimation for making it more secure. But Teac has decided to risk here, and has released 12/10/32 drives based on FireWire and 8/8/32 - based on USB. And if the first model has 400 MBit\sec FireWire, than the second has 1.5 MBytes/sec USB and 8X (1.2 MBytes/sec) write/rewrite speed and that's really risky.

I should also attract your attention to Mobile DVD/CDRW from Addonics. The title speaks for itself, and taking into account, that high scale of integration is appreciated in the mobile configurations, such variant has quite a right to live.

New 16/10/40 models continue to advance - i.e. TDK is preparing some drives based on a chassis from Sanyo that seems to stand better and better on position of similar solutions supplier. There are also appropriate disks: Traxdata has announced some, still only 650 CD-RW, but already rewriting with speed from 6X to 10X - a whole disk will be copied for not more than 7 minutes.

As on any Comdex, the dispute about re-recordable DVD formats has proceeded. And it's late to argue about DVD-RAM. We've only to look, for example, at introduced by Panasonic second generation DVD-RAM solutions - 4.7/9.4 GBytes disks and drives, that cost less than first generation 2.6/5.2 GBytes DVD-RAM at the moment of their appearance. Panasonic drives will cost about $549.

Pioneer on the same Comdex has presented the first drive capable to write almost any DVD disk. Its CD/DVD recorder knows how to write: 1X/2X DVD-R (it's also the first drive that knows how to write 2X DVD-R), 1X DVD-RW, 4X/8X CD-R and 4X CD-RW. It reads just anything inserted into it. However, there's still more of the prototype than of the commercial product. And its market looks very tiny.

DVD+RW admirers (and DVD-RAM opponents) again have tried to use this Comdex as a tribune - three of six companies supporting this standard (HP, Philips and Ricoh) have shown some prototypes of the drives, having visually proved their compatibility - DVD+RW disk could be replayed on a usual DVD videoplayer and DVD-ROM drive. But DVD-RAM already has such a good strong position in the field of re-recordable DVDs, that there are few to none chances to overthrow it from his pedestal.

But we'll see what happens next. Let the time pass and the prices fall. According to analysts, prices for DVD-ROM drives should fall approximately by 15% in this quarter, and for CD-RWs - by 20%. Not much demand for PCs, slow model replacement with fast ones, and we get a very nice result.

Sony hasn't proved itself in the field of optical drives, having preferred magneto-optics in November and having presented its 5th generation drives (and disks). The size, comparing to the previous generation, has doubled (and it made 14X times more comparing to the 1st generation - from 650 MBytes up to 9.1 GBytes on the disk). In addition, the bandwidth has grown by 20% comparing to the previous generation: stable transfer rate makes 6 MBytes/sec, peak rate - 20 MBytes/sec. However, comparing to DVD-RAM 2.0 all this looks bad and prices of $2,600 for the drive are just awful. However, magneto-optics market has always been standing aside the mainstream so to say.

And there's one more market for which Sony prepares its new products - optical data libraries. The prototype based on blue laser has been shown already October, having about 120 GBytes per usual 5.25" disks. It's interesting that the commercial production will begin only this spring and the brands such as HP and Plasmon has already stated complete support - they have announced '2002 year releases of 10-20 Tbyte library racks.

For those who need to store smaller data volumes, Seagate and Imation have simultaneously released new streamers of Travan format. And both companies have presented only external Plug&Play models - Seagate with USB interface, and Imation - with FireWire. Taking into account the size of Travan cartridges of up to 20 GBytes, the Imation's approach looks more nicely, though and it's not so effective as it could be, having in mind interface difference: 5.1 and 7.2 Gbyte/hour accordingly to the first and second case.

But HDDs have such volumes copied instantly. Especially such as new Cheetah from Seagate: 63.9 MBytes/sec would give us 225 GBytes/hour. So, in the 1st quarter of '2001 there are two new Cheetah models released - 36XL and 73LP. Actually, it almost the same line, first is ends with 36 GBytes - $615, and the second starts from it. The most preferable interface for both models is 2 GBit/sec FiberChannel (and 73LP can also support Ultra160 SCSI).

And for those huge size maniacs a new monster, Barracuda 180, has been presented. Undoubtedly, the size is biggest in the world. But certainly for the appropriate price of $2,195. However, taking into account its size, we have absolutely normal HDD price/mbyte rate of $0.012. This drive has Ultra160 SCSI interface and it's fairly enough for 7,200 rpm Barracuda. It's pity, that XBox wouldn't have such HDDs - they'd look very exotically.

So, now we have moved to the consumer class from High-End via XBox. And here Maxtor and IBM have competed this month offering 60 GBytes 7,200rpm models. DiamondMax Plus 60 that has begun to be shipped in November is $50 cheaper than appearing only in December Deskstar 60GXP. $300 versus $355 for elder models.

Notebook owners, have as usual been pleased only by Fujitsu, that has announced a line of 2.5" HDDs based on 10 GBytes disks with total sizes from 10 to 30 GBytes. More features: 4,200 rpm, 12ms search time, 28.7 MBytes/hour bandwidth, UltraATA/66 (it's urgency in notebooks can be fairly doubted). However, some may prefer Fujitsu's internal hard disks to QPS's external ones. Having quite mobile dimensions of about a palm they have the same size of 30 GBytes but with FireWire instead (or 400 MBit/sec and hot connection) and a possibility of stacking up to 5 HDDs in one block.

Let's talk more about interfaces. We have already noted Cheetah's 2 GBit/sec FiberChannel. In November Adaptec has announced the first AFC 9210LP controller for this interface. And that makes SCSI look puny - 2 GBit/sec in case of a duplex operation gives us 400 MBytes/sec bandwidth. Though, comparing to 320 MBytes/sec Ultra320 SCSI, the difference nevertheless isn't so great, and besides this standard has its advantages. By the way, in November its support has been announced by the same Adaptec together with Seagate and LSI, so 2 GBit/sec FiberChannel and Ultra320 SCSI controllers should appear on markets already in summer. And another big RAID controller manufacturer, Mylex, has also announced on Comdex its support of both of these standards.

But it's still High-End. And moving to some "lower ends" I should mark the release of a draft Serial ATA v1.0 specification variant. Though it's not so much of Low-End because even the first variant, defined in the specification - Ultra SATA/1500 - has a bandwidth of 1.5 GBit/sec and that's more than that of today's 1 GBit/sec Fibre Channel! And there are also 2X and 4X variants ahead with bandwidths up to 6 GBit/sec. Well, if Serial ATA is announced to be IDE's successor, it must reside seriously and for a long time.

However, Serial ATA is far from real today, but we'll see USB 2.0 South Bridge already next year. And its 480 MBit/sec will quite suffice to connect even ATA/100 HDDs. It's obvious that QPS is already preparing some variant of M3 UltraSlim Hard Drive with USB 2.0 interface. They promise a peak speed of 50 MBytes/sec and it's more than enough for any task, including "killing" FireWire as a similar devices interface.


Write a comment below. No registration needed!


Article navigation:



blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook


Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.