ABIT KR7A-RAID + Athlon XP: explosive combination
Athlon XP: it is the temperature, idiot!
... Paraphrasing the brutal one, but realist, comment of the North American politics: "it is the economy [what it imports], idiot!", of so bad memory for Mr. George Bush (father) in his frustrated electoral struggle with Mr. Bill Clinton.
The Athlon XP bases on the recent nucleus Young pigeon, the same one on which there are based the Athlon MP (I supported guaranteed multiprocessor) and the Athlon 4 (for portable). It has diverse internal progress in the ambience of the yield, as well as support of the instructions SSE (developed originally by Intel for his Pentium III), but one of his most significant progress is a minor heat generation.
And the saving is not any prank: 15 % less! This has allowed to AMD to keep on increasing the speed of his mikes, being supported for the time being in a technology of 0,18 micron s manufacture . Of course, in a few months (perhaps 4, 5 ó 6 months?) inevitably they will have to go on to themselves to 0:13 microns, but meanwhile they have a breather and he avoids to have to install turbojets instead of "simple" fans.
Certainly, the reputation of "warm" of the mikes AMD was deserved when it was a question of comparing the original Athlon with Pentium III, but it it was not so much on having compared it with the Pentium 4. Nevertheless, the 0,13 micron s new Pentium 4 (model of 2,0 GHz "A" and Superiors) do not consume any more than 55 W, he codes one much more manageable; luck has AMD of which right now they cost more than 360 $ (approximately 410 euros, neither more nor less).
To improve these thermal characteristics there has changed also the material of the encapsulated one of the mike, which has gone on from the bluish ceramic material acquaintance to an orange organic, well brown material or, as in the mike that we have proved, green.
As for the difference of yield between an Athlon and an Athlon XP, it is interesting although not spectacular: between 1 % and 10 % of increase, with 5 % as typical number and more than 10 % if the application is hard optimized for the instructions SSE and does not have scarcely support for 3DNow! of AMD.
Ah, let's not forget the matter of the nomenclature of "equivalent yield with the Pentium 4". The Athlon XP 1600 + that we have proved works really to 1400 MHz, but in the majority of applications it gives good results so much or more than a Pentium 4 to 1600 MHz (1,6 GHz); this politics will be able to please or not, but AMD has been very conservative on having done the equivalence and the tests demonstrate that it has the whole reason.
Of course, unfortunately the Athlon XP cannot settle in any badge with Socket A: the circuitry must be prepared for this mike, what implies that many of the first badges Socket A does not support it. Also, the BIOS must be also ready, although normally a simple update will be enough. If his badge has half a year or more, the best thing is that it consults with the manufacturer from what model of badge and of what review of the same one (what version of the "PCB") it is supported or not.
RAID and ATA/133
ABIT is still faithful to his politics to do a version with RAID of all his basic badges, and follows also faithfully the checkers of the company HighPoint. In this case we meet the most modern model of this company, the chip HPT372, which offers us:
On the RAID in badges ABIT with checkers HighPoint we have already spoken sometimes previous, and really as soon as into many aspects as the aspect of the BIOS RAID it has not changed practically anything; we send you to the corresponding paragraph of the article about the ABIT KT7-RAID.
In the positive aspect, practically quite: good speed with or without RAID, very stable, with drivers available for Windows XP and Linux (if his distribution does not include them, it will find them in the HighPoint web site)... Nevertheless, we have found a small mistake: if an alone CDROM gets connected in a channel (like "teacher" without "slave"), the checker HPT372 was not managing to detect it.
It is not anything very seriously, of course: we can place it in any of 4 positions that the chipset controls ROUTE or in the HighPoint like slave. After everything, it turns out to be a waste to put a pot as slow as a CDROM (approximately 8 MB/s... and thank you) in a channel that can reach up to 133 MB/s; that in accordance with 100 MB/s of the chipset ROUTE.
As for the support of ATA/133 or UltraDMA/133, the first hard disks that continue this standard are coming right now to the shops. We still have not managed to prove them, but do not wait for a big yield increase; since we saw some time ago, even a hard disk of 7.200 rpm of upper intermediate scale hardly overcomes internally 35 ó 40 MB/s, so the "neck of bottle" is inside the disc.
If to use one of these discs like UltraDMA/66 instead of as UltraDMA/100 he supposed like much 2 % of progress of the yield, it is not of hoping that THEORETICAL 133 MB/s should take advantage much better, except on discs of the highest scale 10.000 rpm, in applications that fit in the cache memory of the disc (approximately 2 MB or less) or perhaps in arrays of discs in way RAID 0. As soon as it is a question of reading more than 2 followed MB or of reading something for the first time (as on having loaded a program,), we will meet the limit inherent in the mechanical nature of the hard disks.