by Andy Yaschenko
02/23/2004 | 06:37 PM
Let’s take the companies in the “weight” order: Intel will be discussed last and AMD just before it. Traditional outsiders, Transmeta and VIA, come first. Transmeta summed up the results of the last quarter and found itself losing $21.6 million with a sales volume of $3.6 million. This should come as no surprise, though, as the company has been constantly losing money (a loss of $72 million in 2003). They still hope to become profitable in 2005 with their 90nm Efficeon. A couple of new processors may help, too. The company targets its Crusoe TM5700/TM5900 in an ultra-compact packaging for 1U servers, printers, points-of-sale and everything else, but never mentions notebooks! Well, no one tells you can’t use those processors there. Anyway, Transmeta is groping for a new niche to apply itself to and that’s good for the company.
<%BANNER[article]%>VIA Technologies feels all right as it is, but it is also waiting for the 90nm Esther core (2GHz chips) to be produced by IBM (poor TSMC! VIA left it, too) since the second half of this year, the same timetable as we have with the Efficeon. Besides the frequency, there’s scanty info about the new chip, and I don’t feel like making wild guesses. Let them release it first (if it ever be released). However, VIA has every reason to be optimistic when it promises to ship 5.5 million processors this year (against 3.6 million in 2003, although they had planned this number to be as high as 6 million).
These numbers are negligible for AMD as it sells more chips in a single quarter. This quarter they are planning to sell more than a million of Athlon 64 CPUs; with their appealing price, this plan has every chance to be a success. Meanwhile, the same low price makes dubious the realization of the other declared goal – to end the quarter with a profit. We’ll talk about the monetary predicaments of AMD shortly. Right now – processors.
In January we saw the Athlon 64 3400+ arriving on the classic ClawHammer core, with 1MB of L2 cache and a frequency of 2.2GHz. While the official volume price is $417, first exclusive samples emerged in super-expensive Japan at $437 only. By the way, the official price of the previous flagman, the Athlon 64 3200+, changed considerably. AMD dropped it from $417 to $278. The other 3200+ processor (Athlon XP) went down in price, too, from $325 to $213.
The Athlon 64 3400+ showed up in two slightly different incarnations: for desktop systems and as the Desktop Replacement (DTR), that is, for big and hungry notebooks. That makes the new processor series from AMD the more valuable: Mobile Athlon 64 2800, 3000+ and 3200+ are positioned as truly mobile CPUs. Their price is just a little higher than that of their desktop counterparts; well, the same applies for their power-saving capabilities.
New products are also looming on the horizon. The new CG revision of the Athlon 64 should see the light of day sometime next quarter. Such processors should be less fastidious about the memory you use and, theoretically, should work faster with this memory in dual-channel mode, as the CG core will be the one used in future 939-pin CPUs.
I have no doubts this year is going to be financially successful for AMD exactly because the Athlon 64 has grown mature and strong. The quantity of shipped processors of this family was relatively small in Q4, but the results were anyway very pleasant: a sales volume growth of 76% over the previous year and of 26% over the previous quarter. Add a profit of $43 million, too!
Of course, one successful quarter couldn’t wipe out the losses of the three previous ones, so the company reported a total loss of $274 million in 2003. Well, this is even better than the $1.3 billion loss in 2002! As I mentioned above, AMD hopes to be profitable in this quarter, too, by increasing the share of the Athlon 64 in their production although the start of each year is traditionally an unfavorable time for all vendors.
January, Intel also summed up the results of Q4 and 2003. The numbers are incomparable and astonishing. They notched another record this quarter: $8.74 million sales (+12% over the previous quarter) and a net profit of $2.2 billion (31% higher than in Q3 and twice higher than in the Q4 of 2002). Intel raked in $5.6 billion in 2003, having sold its products for $30.1 billion. Just like AMD, the company is all enthusiastic about the next year.
The enthusiasm is well grounded. First of all, the Prescott made his official debut at last. Well, they started taking orders for this processor in Europe in middle of January already. However, there are serious doubts about the Prescott’s being able to improve Intel’s positions this year. High-speed Prescott-core processors are associated with the Socket T (the 3.6GHz Pentium 4 will support only this socket), and LGA775 processors will constitute less than 50% of all processors sold by Intel in 2004, and their share will only be 8% in the second quarter.
The frequency growth, as scheduled, is small, too. It is going to step up from the original 3.6GHz in the second quarter to 3.8GHz in Q3. It’s probable that they can drive the clock-rate to 4GHz by the end of this year, but I won’t be too sure about it. We should also keep it in mind that the Intel Pentium 4 E “Prescott” chips provide a bit worse performance than the Intel Pentium 4 “Northwood” processors of the same frequency due to the redesigned architecture; therefore, the 90nm Pentium 4 E will really get faster than the 130nm Pentium 4 when 3.6GHz and faster flavours are unveiled.
Curiously, as it turned out in January, Intel will launch a 3.4GHz Pentium 4 Extreme Edition in the LGA775 form-factor in Q2 2004, which was initially expected to be an exclusive prerogative of the Socket 478. It’s pretty evident that this processor on a 0.13 micron Gallatin core will boast a higher performance than the new 3.4GHz Pentium 4 E on the new core because of larger cache. Moreover, it may turn out that the chip will be even faster than the Pentium 4 E processor at 3.6GHz also launching in Q2 2004, as if the Prescott 3.6GHz was speedier than the Extreme Edition at 3.4GHz, why would Intel launch a $999 chip that is slower than a $637 one?
The new Intel’s core for the Pentium family processor, code-named Tejas, is also rumored to have some problems. Sources say it is much hotter than the Prescott, which is far from being a cool guy itself, to put it mildly. The Tejas is said to delay from the Q1 of 2005 (where it was postponed from the Q4 of 2004) to the Q2. This pause will be filled by the arrival of the 4.2GHz version of the Prescott. Well, it would be a nice opportunity for AMD if AMD meets its schedule for once and releases the 90nm Athlon 64/Opteron chips in time. Those rumors also suggest that the server counterpart of the Tejas, the Jayhawk, is also postponed to the Q2 of 2005, but this is actually one and the same core.
The 90nm Celeron will hopefully have a better fate than its full-fledged Pentium 4 E mates, considering the deficit of such value processors. Intel is trying hard to promote its high-performance Pentium solutions in the first hand, although the market seems to have had them enough, while the demand for low-cost processors with high frequency and exciting performance is always quite high.
So it is the more pleasant that the first 90nm Celerons are coming along with the 90nm Pentium 4 in early February, with frequencies ranging from 2.53GHz to 2.8GHz and a doubled (256KB) L2 cache. Overall, 533MHz processor system bus and 256KB of L2 cache are both appealing options, and Intel will have to stifle it once again.
As for the upcoming server processors, another representative of the 90nm Prescott generation is going to rise up in the near future. This spring, the 90nm Xeon on the Nocona core comes to us with 1MB of L2 cache and (first ever on Xeon!) the 800MHz FSB. This cache size is just the starting point, I guess.
I hope the energy consumption of chips on the Nocona core will remain acceptable. Otherwise, I’m afraid just to think of the power consumption (and heat dissipation) of the 64-processor server on Xeons IBM is promising to roll out at the beginning of the next year. Curiously enough, this application of the Xeon is not quite in accordance with Intel’s own ideology, which places only the Itanium at the high end.
Well, last year Intel sold about 100 thousand Itanium 2 chips – not bad for processors designed for high-end servers. But the Itanium has been on the market for more than 3 years now, no surprise that its sales are picking up these days. However, there is something Intel Itanium 2 lacks and that holds customers from acquiring the pretty expensive hardware and need for investing in software – poor performance in 32-bit mode. In the middle of January the company unveiled an IA-32 Execution Layer (EL) to work above the operating system and to raise the efficiency of Itanuim’s execution of 32-bit code by 50-70% (Intel’s estimates). Anyway, x86 IA-32 instructions are converted into IA-64 ones on the software level, with the unavoidable performance degeneration.
That’s why the most important episode of the last month was the rumor that Intel would give up completely and show the Yamhill technology at the February IDF – their own 64 bits for x86 processors.
With all its uncertainty, the Prescott processor was in the focus of the previous section. Let’s continue in this manner. Well, there has been no revolution here in mainboards: manufacturers were all announcing slightly redesigned versions of their older products. An updated BIOS plus enforced power circuitry usually do the trick. If the old product already boasts a strong power unit, they just declare it supports Prescott. The second way is rather more economical, but the point remains the same.
First incarnations of the core will plug into Socket 478 mainboards on different chipsets: the duo from VIA and SiS, the PT880 and 655TX, respectively, try to compete with the i865PE (and even i875P) with some success. The chipset from SiS looks more appealing, but they both are cheaper than Intel’s chips. You can already find mainboards on these two chipsets in stores.
This is what we have now, in other words, what is going to become our past pretty soon. Those who are looking ahead see a new generation of chipsets on the horizon. It is as uncertain as the Prescott, frankly speaking. DDR2 is unlikely to provide a perceptible performance boost over ordinary DDR: it will be more like a statistical error in many benchmarks. The PCI Express bus would be more interesting for the end-user, but it is not going to show its full potential in this year, considering the attitude of the card makers.
However, these reflections shrivel before the flourish of trumpets from the marketing departments of the chipset makers: Grantsdale and Alderwood come out this spring. We should learn their official denominations: i915 and i925, with some derivatives. Some sources say that the i925 will come with a letter “X” rather than “P”, while the i915 will come in three versions: P, G and GV. The last version, GL, is supposed to come out as 910GL. To all probability, i915GV and i910GL will appear in the second half of the year only, just like it was with the 865 series.
The direct rival of the i915P from VIA, the PT890 chipset, should come out around that time, too. At least, engineering samples’ shipments should have been started in the end of January. So, very soon there should be mainboards on this chipset already and they will come into the market no later than i915P-based products. This is the direct outcome of the reconciliation with Intel, which has started bringing fruit. SiS is somewhat tardy: samples of its SiS656 go to mainboard makers in February, while its production will start in April only.
Mainboards on integrated chipsets won’t bring us anything new in the near future. Only the RADEON 9100 IGP, which has been around for a while, has been steadily gaining its ground. In January, two mainboards on this chipset appeared in shops (Gigabyte GA-8TRS300M and ASUS P4R800-V Deluxe), and the RS3M from MSI was announced. Interestingly, the mainboard from ASUS claims official support of the Prescott.
We should wait for a while yet for the new generation of integrated chipsets to appear. The 915G will show up closer to the summer, while the 915GV and 910GL will arrive in the second half of the year only. The 910GL will support Socket 478, so today’s Celerons shouldn’t be dismissed altogether. Future Celerons will also work with this chipset as it promises support of FSB533.
The competitors are making ready with their answers. ATI will present its 400 series in the second half of the year, too, when the full-fledged RS400 comes out with a DirectX 9 core and support of dual-channel DDR/DDR2 and PCI Express. Its cut-down versions are called RC400 and RU400. The former will have one of the memory channels disabled, while the latter will also lack support of PCI Express. VIA Technologies is planning to release an integrated version of the PT890 this year. The PM890 will feature VIA’s own graphics core derived from the DeltaChrome. SiS is slow again: the SiS662, an integrated version of the SiS656, will be sampled in the summer and will come out in the fall.
There’s one more thing about integrated chipsets. It has now become popular to develop a miniature mainboard on such a chipset and install, or even solder up, this or that mobile processor. In January we saw the ECS L7VMM3, which is a combination of the KM266 and the mobile Athlon XP 1400+, and the EPoX IP-4MTS2B (Pentium M/i855GM).
A special product in this area comes from Freetech: mainboards on integrated chipsets (i865G and i845GV) for ordinary processors (including Prescott). The manufacturer employed VIA’s etBIOS technology to play audio and video discs without booting the OS. The idea is exciting, although it doesn’t seem quite to the point – you won’t get less noise anyway.
Now let’s turn to solutions for AMD processors. They were scanty in January, save for a few mainboards on the nForce3 250 chipset that were demonstrated by AOpen, Chaintech and DFI at CES. The companies belong to the middle class and there’s no one from the “top four” to showcase the nForce3 250. No name from the top four and overall there are few names. Well, that’s quite understandable, since there is the K8T800 that has been supporting Serial ATA-150 for long.
It starts looking as if the things that happened in the last year were VIA’s misfortune rather than NVIDIA’s success. And this is quite a sudden turn of the plot. At least, this is what the persisting problems NVIDIA has with its South Bridges imply. By the way, South Bridges may become a hot topic in this year. First, each chipset for the Athlon 64 is in fact a South Bridge. Second, DDR2 won’t be a real thing in 2004, so South Bridges may become the crucial factor in evaluating chipsets for the Pentium 4 platform.
That’s where SiS comes onto the scene. In the middle of January they announced their SiS965L, their first South Bridge with support of PCI Express, 7.1 audio and compatible with the majority of North Bridges the company already has, ranging from SiS648FX to SiS760. There’s only one problem – there’re no PCI Express cards announced yet (I don’t count in graphics card that don’t support x1 PCI Express ports anyway). This chip will come into mass production no sooner than May. Its successor, the SiS966, with support of Intel High Definition Audio (Azalia) will appear in the fall when the integrated SiS662 shows up.
VIA Technologies is already showcasing its new South Bridge, the VT8251 chip. It looks quite weird: up to 8 Serial ATA devices, Gigabit Ethernet, 7.1 High Definition Audio, PCI Express. Mass production of such chips is expected to start next quarter. In fact, this is reasonable considering the vague situation with PCI Express cards. But after the marketing folk from Intel start their propaganda, the market may succumb. Sometimes they can do their job terribly good.
Anyway, VIA is boiling with optimism. They sold about 31-33 million chipsets last year, and are going to push this number as high as 45 million in 2004 to become the leader of the market. And this optimism comes after they lost 19% of their market share in 2003. However, this time the experiment is pure as VIA’s chipsets are “pure” legally.
The December results were not wonderful for the top four companies as sales slumped by 14-20% compared to November (this is largely due to the December holidays, though). As for the entire year, only ECS found itself completely on the losing side having lost 69% of its market share compared to 2002. The rest of the companies enlarged their shares: Gigabyte by 4%, ASUS by 12% and MSI by 20%.
They are all optimistic, too. All four are promising sales growths of 10-35% in 2004 (MSI and Gigabyte are more prudent in their estimates), while ASUS (with ASRock) and ECS promise more than 30%. Gigabyte hopes for 10% and I have no doubt they’ll make this. By the way, in February-March Gigabyte is going to announce its new trademark “Gigatrend”, which is in fact analogous to ASUS’ ASRock.
At last! The price of 256Mb DDR400 grew by 10-12% in the first two weeks of January. The demand was really on the rise, however slightly, due to several reasons: mainboards with dual-channel DDR SDRAM had been selling well and the manufacturers were willing to restock for the long celebration of the new Chinese year.
What was more surprising, the prices went up again after the holidays. Overall, they jumped from $3.7 to $4.65 for one 256Mb DDR400 chip. It’s quite an unrewarding job to try to make predictions about the prices of the spot market now. We just see that the open market of DDR has livened up somewhat, and the rest of the story will be in February.
A dead calm is in the market of extreme modules. Well, Corsair acted up with its 2GB module kits (registered, for the Athlon 64 FX) and PC3700 registered modules for the same processor. 500-533MHz modules from Kingston and OCZ are not thrilling anymore, while OCZ also released its PC-3200 Platinum Limited Edition EL DDR with 2-2-2-7 timings. I hope someone will feel the difference.
Feeling the difference is also the problem with DDR2 that’s getting closer to us. In January Elpida started mass production of 2GB registered PC2-4300 DDR SDRAM modules based on 533MHz 512Mb chips. This is running ahead of the earlier announced schedule, so Intel’s money hasn’t been wasted. Samsung also announced a 2GB registered module in January along with unbuffered 1GB module. Samsung has versions for all speeds: PC2-3200, PC2-4300 and even PC2-5300 (this is only for the ordinary 1GB module), but they only talk about sampling their chips.
Well, it’s better getting along cautiously, especially as the perspectives of DDR2 are rather vague in this year. Manufacturers of memory chips, mainboard makers and analysts are getting more skeptical about DDR2 being able to become the leading memory type in the market in 2004. It’s more probable that the difference in price of DDR and DDR2 won’t be at first proportional to their performance ratio. Although there are numerous module samples, many manufacturers agree that they will start mass production in Q3-Q4 of this year, or in early 2005.
So far the manufacturers have been busy trying to reduce the production cost of a DDR2 chip – by increasing their capital investments. Korean DRAM manufacturers alone are going to invest about $6 billion into modernization of their facilities, expanding them and transferring them to 300mm and tech processes 110nm or 90nm. Nanya is going to have 40-50% of its facilities using the 110nm tech process by the third quarter of this year. To produce mainly DDR2, of course. Micron also increases its capital spending to $1.3-1.6 billion. Overall, they are all investing money with rather hazy perspectives of ever returning them.
It will be really interesting to know the changes in the companies’ ratings by the end of this year. I guess some changes are sure to occur (one thing is only certain – Samsung’s first place). At least, the beginning of the new year has brought a small sensation as Infineon lost its third place to the invincible Hynix by the Q4 results. Hynix has now 15.8% of the world DRAM market against Infineon’s 14.6.
It was largely due to the strong positions of Hynix in Asia, China and Japan, but can it keep them in the future? For example, the Japanese government under the pressure from Elpida is going to discuss introduction of protective duties against the Hynix produce. The rate of duties is as high as 40%. In other words, such duties close this market for Hynix completely.
The only country where this trick cannot pass is China. Hynix actually has its own production there: chips are produced by the company’s partners. However, Micron is going to enter this market, being enticed by the rapid growth of the Asian market, so Hynix should be on guard.
People from ATI Technologies rejoice. According to Mercury Research data for Q4, ATI has outdone NVIDIA in the market share (for the first time in years!) The fact is meaningful, although the gap itself is small (ATI’s 24.9% against NVIDIA’s 24.7%). Having lost 3.2% in the last quarter, Intel still remains an unrivalled leader with its 31.7%.
Anyway, NVIDIA is in the third place now, and this is not due to the failures in the notebook sector (or, rather, due to the success of the RADEON MOBILITY 9600 – it’s all relative in this world), but because of its losing in the field where it has always been a traditional leader! ATI is selling ever more GPUs for desktop graphics cards, while NVIDIA’s sales are declining.
The technical capabilities of the products are one thing, but ATI is also aided by its own suddenly found talent to work with partners, while NVIDIA seems to have lost it. The story with ABIT is quite characteristic: they settled up somehow on mainboards, but didn’t on graphics cards. Meanwhile, ATI agreed on collaboration with ABIT in January and we’re waiting for a full range of graphics cards on the RADEON chips. And this was not the only acquisition of ATI in January, although ABIT alone would be quite enough. Terratec confirmed the rumors by officially announcing their union with the Canadians, while the Korean Sigmacom also improved its partnership with ATI and Soltek decided to add some RADEONs into their product list. A kind of compensation for NVIDIA: EPoX decided to start a life of a graphics card maker with GeForces.
Enough of theory, let’s have some product news. As the intermediary season between chip generations continues, there are few new products, actually. It’s all calm and quiet in the low-end sector, with some more activity towards the mainstream. Soltek launched a couple of cards on the GeForce FX 5700, and Transcend reacted to this with a standard RADEON 9600 XT.
However, there’s some news here. Club 3D and PowerColor are struggling hard to push up the S3 DeltaChrome S8 by offering another, quite nice graphics card on this chip, called SS8-D3L, with passive cooling and costing just above $150 for a 256MB version.
The main events happened in the high-end sector, as usual. The single card from Gigabyte on the GeForce FX 5950 Ultra is no great event, we’re more interested in future perspectives: NV40 and R420 are quite close. The NV40 is rumored to hit the stores in late spring (May – April), called something like GeForce FX 6xxx.
We’ll surely see prototypes of graphics cards on this 0.13-micron chip at CeBIT, and that means we’ll see performance tests of this monster made of 175 million of transistors. The supported interface is AGP 8x, since NVIDIA seems to have problems with transferring to PCI Express. The company preferred to make the new series with AGP, but with an option of using an external AGP-PCI Express bridge. It’s clear the capabilities of the spring solutions from NVIDIA won’t exceed the AGP 8x limitations due to physical reasons (and that’s yet another reason for not being hasty with the i915/925).
However, it’s doubtful the PCI Express bus will give more performance to the R420, although it should have a full-fledged realization of this interface. It’s just not clear if the R420 needs more bandwidth to the CPU. We’ll learn this shortly, as ATI is rumored to be a bit ahead of NVIDIA in chip revisions. So we’re looking forward to CeBIT.
As for the XG45, the DirectX 9.1 successor to the Volari V8 Ultra, we’re not likely to see it at all. Yes, the chip was going to be announced in March (well, they may announce it – it’s no hard matter), but then the announcement was postponed to Q3. In other words, XGI completely skips over this year, and graphics cards based on the new chip will arrive no sooner than by Christmas. Maybe I’m wrong and the company will be doing all right, but I feel like we’ve got a graphical Transmeta, who regularly shows some signs of life, but nothing more.
As usual, the junior model introduced in January had a diagonal of 17 inches. Well, it was actually the single model of the series. Sony announced its SDM-HS73P, a 17” LCD monitor with a proprietary coating of the screen called Onyx-black that should provide for an excellent color reproduction, brightness and contrast. The brightness of 400nit is really good for a 17” box, but not extraordinary. Add also 16msec response time and 160° viewing angles.
I would also like to single out the third modification of the 17” model BenQ FP767 where the manufacturer emphasizes another parameter: the pixel response time is as low as 12msec, which is the best responsiveness among LCD monitors of today. The rest of the characteristics are above the average: 500:1 contrast ratio, 300nit brightness, 0.264mm point.
The parameters of upcoming monitors from Xerox will be around those, too. The company gives its brand, but won’t produce anything – they’ll give it all over to the Taiwanese. Proview Technology is rumored to have drawn the lucky number – the order should be hefty enough. 13 LCD monitors under the Xerox brand are going to be priced starting at $399. The price is expected to go down in the second half of the year.
At least, DisplayResearch promises a total price collapse: 17 and 19” LCD monitors will be getting cheaper starting from the second-third quarter when 5-th generation production lines start working. So the price for panels will be reducing by $3-5 monthly. Sale prices for 15” panels, according to DisplayResearch, would remain the same at best, as the demand for them still exceeds supply.
The G5 production lines should say their word in this year. The manufacturers hope to increase their shipments by 40-90% per year, in fact, by a half. And that’s only the beginning. The amount of capital investment in this area will be as high as $9.42 million in 2004! They have already started transferring to equipment for making 6G wafers.
Samsung is among the first: the $2.3 billion chunk of the total of $6.7 the company is going to invest into the production this year is intended for building and equipping its 6G fab, which is going to start working in the second half of 2005. The factory will be mostly concerned with LCD TV-sets.
Such devices are becoming ever more popular. Sharp, the leading manufacturer of LCD TV-sets today, plans to triple its production volume at the Japan fab by adding two new lines to the existing one. Accordingly, the production volume will grow from the current 15 thousand to 45 thousand panels per year. It’s more than half a million TV-sets with a diagonal of 26”.
In January, Sharp also announced six new “Aquos” LCD TV-set models, assembled from 6G panels. These devices with a diagonal of 26-37” and a price of $4,000-6,600 will start selling in February-March. A 45” model will complete the series in May.
However, the diagonal size is not the only thing that matters. In January, Sony released a cute product called LF-X1, which is a portable 12.1” LCD screen (I just can’t call it a monitor or a TV-set). It comes with a dock station with an integrated TV-tuner and a Wi-Fi transmitter. The screen contains a receiver that can reproduce the MPEG4 video stream broadcast by the station. In other words, we’ve got a TV-set, which looks like a tablet.
On the other hand, the 12.1” diagonal is rather small, especially if you like plasma screens. While the 61” PX-61XM2P from NEC is normal enough, the 80” panel with 1920x1080 resolution showcased by Samsung is a kind of record (LG’s 76” screen was the previous record-holder). This is only a sample, though. Such screens will start shipping from the second half of the year when the appropriate production line starts working.
Of course, you can’t go around with a plasma panel in your pocket. There’s a solution from Leadtek (Leadtek is selling what eMagin Corporation produces, to be exact) – the Leadtek X-eye. The OLED matrix provides an acceptable image quality equivalent to watching a 50” screen from a distance of 1.5m. The resolution is 800x600, power goes though the USB cable, and the device connects across the standard analog interface.
Another exotic topic of the month was electronic paper. This idea emerges from time to time, but was quite actively discussed in January. Philips announced the foundation of Polymer Vision to produce flexible screens: 5000 items annually with resolution of 320x240 and a diagonal of 4.7 inches. Fujitsu Laboratories showed a nice concept with brightness comparable to that of ordinary paper. Fujitsu doesn’t promise this technology to come into the market before 2006, though.
It seems like ordinary 3.5” hard disk drives for desktop PCs have finished their evolution. Well really, what can you offer to the user after those 250MB models? 300GB? But what for?
That’s why there is no news in this area at all. Capacities remained the same, they only discuss interfaces. First versions of Serial ATA II Gen1x and Gen2x specifications will be shown at the February IDF, where “x” denotes external devices. That is, external drives will connect to the computer across Serial ATA as they do across FireWire and USB 2.0 now. Theoretically Serial ATA may oust the competitor interfaces due to its higher bandwidth (1.5 and 3Gb/s, respectively). External Serial ATA drives are not perfect, though, as there’s no unification in connectors and power with the internal Serial ATA HDDs and they also need to use an external power supply.
The speed is not that high as it seems, too. At least, even the 3Gb/s Serial ATA II is no record-breaker, as Hitachi has already reached 4Gb/s with their FCAL against the standard 2Gb/s for this interface. So the race continues, although FCAL is exclusively a sever technology, unlike Serial ATA.
The untiring Fujitsu was prominent in January continuing its series of mobile Serial ATA II Phase I HDDs of the 2.5” form-factor. The company announced new devices from this series – the MHT20xxBH line of 40, 60 and 80GB capacities. The parameters are no worse than those of “grownup” devices: 5400rpm, fluid dynamic bearings, 8MB cache. They are expected to come as soon as April.
A curios trend: no news about 3.5” models, some news on 2.5” HDDs and a heap of news about 1” and smaller drives. First, Cornice SE announced a 2GB 1” hard disk drive (the previous model had a storage capacity of 1.5GB), so we’re waiting for the player from iRiver, which is a client of Cornice. Second, Toshiba showcased its 0.85” HDDs at CES 2004; these are the smallest for today, but have capacities of 2-4GB! Shipments are expected to start in the fall. Another prominent figure in this area, the Chinese Magicstore, also plans to ship samples of its own 0.8” drives in the fall.
To end up with hard disk drives, let’s have a glance at their external versions. LaCie surprised everyone with a tremendous capacity of up to 1TB (the secret is simple – several standard HDDs in one casing). Design and accessories are also things to play with: the same LaCie showed external drives designed by Porshe Design, while Western Digital offers a 250GB external HDD with an integrated 8-in-1 card-reader.
Transcend took to producing external hard disk drives and started with a case with the USB 2.0 interface or USB 2.0/FireWire, which may sell by itself or with preinstalled 30-80GB drives. Quite an interesting approach.
Another interesting approach, in the optical drives area, shows Hewlett-Packard. The LightScribe technology seems to be simple: 1. CD-R discs have a non-operational upper surface where you usually write something and 1. The optical drive has a laser. So why can’t the drive burn those writings for you? You should wait, though, for discs with an appropriate sensitive coating and drives that can do this trick. The technology is expected to come around in the next half a year and won’t make the devices or discs much more expensive. At least, the manufacturers of drives and discs are readily licensing this technology.
Now we can turn to optical drives and discs. There were no sensations in January, although some things changed for the better. The 8x dual-format drives from Hitachi-LG, Teac and Pioneer are nothing extraordinary (save for some miraculous noiselessness of the Pioneer DVR-A07-J), but there are interesting rumors. In March, at CeBIT, LG is supposed to showcase a prototype of 12x drive. It’s high time, I’d say. Other players, like AOpen, are ready to follow the suit. In theory, someone may dare to show a 16x model, which are expected into the market by the end of this year.
DVD+RW Alliance also promises to show its standard for 16x DVD+R discs by the end of this year as it was declared at CES 2004. They also showed first dual-layer DVD+R discs of 8.5GB capacity there. The discs and the drives that support them are expected by the market in the spring. It is clear DVD Forum will ratify a specification for 16x DVD-R discs, but DVD-R is becoming less important.
The war is fought for the future. Toshiba showcased their HD-DVD player at CES, quite compatible with current DVD discs, which is a serious advantage over Blu-ray Disc. On the other hand, Hewlett-Packard and Dell joined Blu-ray Disc Founders – that makes them even. It’s going to be an exciting show to watch the monsters fighting to death at the end of this year.
It was very hot in this area, notwithstanding the winter. First of all, it’s not every month that we see both Intel and AMD announcing new mobile processors. This January AMD responded to Intel by releasing three mobile CPUs (Mobile Athlon 64 3200+, 3000+ and 2800+) along with the desktop Athlon 64 3400+; all mobile processor have the same core.
The high ratings imply that notebooks on them will be power-hungry, but high-performing, for systems that replace the desktop PC like the one announced by eMachines: 15.4” LCD panel, Mobile Athlon 64 3000+, 512MB of PC2700, 60GB HDD, MOBILITY RADEON 9600, DVD-ROM/CD-RW, 802.11g/Ethernet 100/V.92 and so on and so forth. Can your desktop PC boast this power? With all this abundance, the price is quite reasonable, $1550.
The response from AMD came after Intel’s launch of the long-anticipated Celeron M, the light version of the Banias. The platform had been good enough, but became simply excellent: 1.2-1.3GHz chips cost about $107-134, while their 800MHz super-economical mate (feeds on 7W only) is evaluated at $161.
This news is altogether pleasant as well as the fact they you’ll soon be able to use those Celerons with cheap chipsets. ATI has announced its signing a license agreement for the Pentium M bus with Intel, which would allow the company to use the current M1683 and CyberALADDiN-P4 chipsets as well as develop new ones. However, there was even louder news: VIA Technologies thinks its cross-license agreement with Intel gives it the right to develop chipsets for the Pentium M and is going to unveil two finished products in February: the single-channel PN800 and the dual-channel PN880, both with the integrated DeltaChrome graphics core.
Good news ends here, while bad news is related to Intel, too. They showcased working notebooks on the Pentium M on the 90nm Dothan core at CES 2004, and everybody was waiting for March when they had been supposed to come out. But just a few days after the presentation it transpired that the shipments of this processor were postponed from Q1 to Q2 as the chip needed some polishing off. You may remember that earlier they had promised it at the end of 2003. The ill-fated 90-nm tech process again…
Interestingly, the growth of complexity and frequency of the new chips fully neutralizes all possible gains in energy consumption. While the current Banias has a heat dissipation of 24.5W, the Dothan should approach 30W by the end of this year. Its successor, the Johan, is rumored to aim at something like 45W. With such big numbers, Transmeta may have its chance to live and even to be successful.
Overall, the future lies in the dark, and the only good news from Intel in January (save for the Celeron M) is the announcement of the 802.11b/g adapter for the Centrino platform. It means new models of notebooks will learn how to connect to wireless networks at a speed of 54Mb/s! You’ll have only one problem – thinking about the use of this bandwidth, save for watching movies across the network.
Well, these are upcoming models, but what do we have today? MSI has made its debut in the market (it was not effective, but meaningful). The company started out by releasing the M3000 system (Centrino, 15” LCD panel, MOBILITY RADEON 9600, integrated 4-in-1 card-reader and so on). Once again, you may think MSI is ashamed of its own child, seeing how it presents every piece of info about it.
As for mature players, they were more interested in the latest Intel’s desktop or desktop-related processors. Leviathans are quite popular among notebook users nowadays. Moreover, gaming notebooks are becoming more and more popular these days. Even corporate-oriented Dell launched its XPS notebook for gamers that integrates Intel Pentium 4 3.40GHz or Intel Pentium 4 Extreme Edition 3.40GHz processors as well as ATI MOBILITY RADEON 9700 graphics MCM with 128MB of memory, 512MB to 2GB of dual-channel PC3200 memory, 60GB (7200rpm) or 80GB (5400rpm) HDD, UXGA Wide-Aspect 15.4” display along with plethora of connectivity features, such as 802.11b,g/802.11a,b,g WLAN controller, Gigabit Ethernet, 4 USB 2.0 ports, FireWire and so on.
The Centrino enjoys popularity among the manufacturers, too. Dell did well releasing two quite different Centrino-based notebooks. First, it is the Latitude D505 (14.1” screen, Celeron M 1.2GHz) for only $1050 and weighing just 2.3kg. Second, it is the more powerful Inspirion 510m with a 14.1” screen, faster processor, more memory and so on. For example, the configuration of a 14.1” LCD panel, Pentium M 1.4GHz, 256MB RAM, 30GB HDD, DVD-ROM costs $1450. Your choice is wide, but the Celeron M-based model looks best among all Centrino notebooks announced in January.
Fujitsu Laboratories made another step to making the notebook user absolutely mobile. The new material for power cells allows making battery elements where 300ml of 30% methanol provide the notebook with power for 8-10 hours. Regrettably, they talk about working prototypes only, and I only hope the idea will have its practical implementation in the near future.
