EK SpeedpluS: overclocking with guarantee
Some tests at supersonic speeds
Really, the system works. We lit it, entered the BIOS (of course, the adjustments of the mike are done without jumper some) in a pair of seconds we were using an Athlon to... 1.200 MHz, or what is the same: 1,2 GHz.
But was already not it a mike of 1,2 GHz? Yes, it is true, the truth is that it is not an overclocking. But after there was the multiplier of the free mike, 100 was not 12 x , but 9 x 133. On having had a badge with control of the multiplier, and of the MHz bus in MHz, the options are almost infinite, so we center on those who were using a bus of 133 MHz (266 "equivalent" MHz), since the final speed of the system is perceptibly major than with a bus of 100 MHz (200 "equivalent" MHz).
(Let's remember that 200 ó 266 MHz of the bus of the Athlon and the Duron are physically the half, but since one makes use of every sign doubly - how in the memoirs DDR - the final yield of the mike is JUST AS if there were "physical" MHz. It is not a trick, it works; the Pentium 4 does the same, making use of the sign 4 times to obtain 400 "equivalent" MHz from 100 physical MHz).
And we keep on rising: 1.266 MHz by means of a multiplier 9,5x; 1.333 MHz by means of one 10x; and 1.400 MHz (1,4 GHz) with a multiplier 10,5x. In this case, already we find certain problems of stability; solution? To raise the voltage of the mike a pair of tenths.
This takes us to another question: in a normal system, to raise the voltage of an Athlon to 1,4 GHz would have put the warm mike as the hell, obtaining hardly the looked stability, but thanks to the special system of refrigeration of the EK SpeedpluS, the mike was scarcely to 30th - 35th. Yes, he has read well: more cold than yourself. Try to obtain this with a normal fan (or with a turbojet), especially in summer...
Very well, next we have the tests; we have used a Duron to 850 MHz and an Athlon to 1 GHz (both with bus to 100 MHz) to see if they are comparable to the whole Athlon to 1,4 GHz with bus to 133 MHz:
Amazing. A few numbers so high that almost they lose sense. To play 1280 x 1024 points as minimum in any game, to activate the anti-aliasing and all other options of visual quality without worrying for the yield... and with the guarantee of that the mike will not burn in any moment.
Nevertheless, the numbers say also another thing to us: this mike is so rapid to 1,4 GHz that the graphic card must be as rapid as possible. Really, if he wants it to play, a GeForce2 MX is not sufficient, it would be saturated completely and would waste the mike. The GTS Pro or Ultra are the only logical partners for this speed... and almost they are small.
3DStudio MAX: the real thing
Oh, yeah... Personally, I understand perfectly the fever that does that someone raises the buy of a team as this one to play to Quake3; to play to "I have more MHz that you" it is possible that it is not very healthy, but it is entertaining.
Nevertheless, the reality is that in most of the applications, including the games, there does not exist so much difference between a mike and other a few tens MHz more rapid. This is due to the fact that many other factors enter consideration: quantity of memory RAM, graphic card, speed of the hard disk...
But there exists a class of applications that depend fundamentally on the speed of the microprocessor: the "professional applications" of hard mathematical base. Between these, we have chosen one of the most well-known: 3DStudio MAX 2.5, an application of drawing and animation in 2D and 3D capable of achieving almost indistinguishable results of the reality... by means of many, many calculations.
Really, the yield of 3DStudio MAX depends on several things, not only on the brute force in MHz of the mike, but also on the quantity of memory (it consumes GREAT), of the band and the latency of the same one (that is to say, of the type of memory), of the breadth of the system bus and of the type of mike in himself. To isolate these factors, we have equipped the system with 384 MB of memory PC133 CAS2 and have made fixed a bus of 133 MHz (266 "equivalent" MHz), changing only the multiplier.
The project MAX has consisted of the renderizado to format of video AVI of one of the animations example of 3DStudio, that of a few dolphins swimming inside the water while the sun is reflected in thin particles. His little more than 3 seconds have been difficult of renderizar, although the result 2,43 MB (file compressed into format ZIP) is really excellent.
Since we see, the progression is almost linear. In EAST I marry yes the investment in more MHz can justify itself entirely; since it sees, thanks to these 400 MHz extra we have happened to realize the work in almost 2/3 you divide of the time... "only 33 minutes instead of 46. How much does he believe that it costs 1 hour of time (or several days) for a company of graphic design? Or for an investigator who calculates for finite elements?
And also, this is almost absolutely independent from the graphic card. A GeForce2 MX would prove more than adapted for this work; think that the previsualizations are not anything so difficult of doing, and at the time of the final calculations it will have only to represent 1 picture for the second or enclosed less.
Undoubtedly, we are not going to say that the EK SpeedpluS should be a team for all the publics. And not because his price is very high, since it moves in the economic stripe, between 200.000 and 250.000 ptas without VAT (approximately 1.100 $ - 1.400 $) with monitor of 17 ".
But it is a little noisy (neither more than other teams with overclocking, and considering his 6 fans it looks like the whole design achievement), it weighs an authentic atrocity (17 kg will not sink his soil, but they will hurt him in the back if he plays at becoming the male with him) and simply it offers much more yield which most of users needs.
In any case, I would say that it has the following applications:
The principal thing is that it is not expensive for what it offers, it works correctly even with overclockings important (although everything has a limit) and can save a lot of time in certain applications. A dual badge can look like a more elegant solution, and sometimes more appropriate, but almost safely it will cost much more.
And it will always be able to change into the future the mike into one with major capacity for the overclocking (at least while I did not change the format Socket A); the theoretical use limit perhaps is in 1.875 MHz (12,5 x 150), which presently is not within reach of a fan, at least in the platform Athlon (certainly, the Pentium 4 is slower enough than the Athlon in 3DStudio MAX 2.5, and much more expensive...).
For more information about the EK SpeedpluS, consult: