Nice one lamsey, I'm glad someone cleared up all the crap that was being sprouted in this thread
Nice one lamsey, I'm glad someone cleared up all the crap that was being sprouted in this thread
Is it really such a good thing ?Originally posted by DWk@7 September 2003 - 14:30
good thing bout p4 is the 800mhz fsb, but i bet getting some ram sticks like that must be reaaaaaaaaly expensive![]()
Once again, you have to remember that higher frequencies means higher temperatures, so it is a far better idea if you can find a way of increasing memory speed without raising frequency. This is the reason for ddr memory - twice the data throughput for the same frequency (DDR400 ram actually only runs at 200MHz). Power consumption rises in proportion to the square of the frequency, so (everything else being equal) memory on an 800MHz bus would generate 4 times the heat of the same memory on a 400MHz bus.
Of course, everything else is not equal, the core memory voltage is lowered in order to reduce the power consumption, this would make lower performance memory unstable so higher performance (and therefore, as you have suggested, more expensive) memory has to be used.
The solution is multi-channel memory transfers, so that more than one memory bank is transferring data at the same time. For example 4 banks of memory each running at 200 MHz (DDR400) could be configured to supply data simultaneously giving an aggregate data rate of 1600 MHz. Power consumption would be exactly the same as a single memory bank of 4 times the size running at 200MHz. Conversely, a single bank running at 800MHz (to give a data rate of 1600MHz) would consume 16 times as much power.
This sort of technology is already in use on high-power mainframe comps, and is now starting to appear in pc chipsets (eg Nvidia NForce2). The real way to progress is to do much more within existing frequencies rather that constantly attempt to produce higher and higher frequencies with all the associated problems.
AMD is showing the way in processor design, it's latest 400MHz FSB XP3000+ processor typically uses less power than the 266MHz FSB XP2000+, and only slightly more power consumption when running flat out - you are certainly not going to get those sort of improvements by constantly raising clock speeds.
Edit: and I wish I had seen Lamsey's piece before I wrote all of that.
.Political correctness is based on the principle that it's possible to pick up a turd by the clean end.
And I learned the above while doing a project in high school for Advanced Higher Computing; it's not difficult.Nice one lamsey, I'm glad someone cleared up all the crap that was being sprouted in this thread
Even the more complex stuff I've learned since then like the differences between the caching methods, the ins and outs of Hyperthreading, etc., are not difficult to understand.
edit: @lynx, the latest AMD Athlon XP is actually the Barton-cored 3200+.
@Lamsey, yes, I know the 3200 is the latest processor, when I refered to the 3000 I was meaning 400MHz FSB was the latest offering. The 3200 processor actually uses more power which is why I didn't quote that one, and there have also been mutterings about the performance of the 3200.
.Political correctness is based on the principle that it's possible to pick up a turd by the clean end.
Ah, sorry, misread you there.Originally posted by lynx@7 September 2003 - 21:45
@Lamsey, yes, I know the 3200 is the latest processor, when I refered to the 3000 I was meaning 400MHz FSB was the latest offering. The 3200 processor actually uses more power which is why I didn't quote that one, and there have also been mutterings about the performance of the 3200.
The 3200+ is probably over-agressively named; I think 3100+ would be more accurate.
What many people don't realise is that the model numbers are based on the original Athlon's performance. The model number reflects the speed in MHz of an original Athlon.
However, the Intel P4 architecture is developing, while the original Athlon is the same as it always was; the Athlon model ratings, which started out better than equivalent P4 clock ratings, are now falling behind their equivalents.
This is because P4s were originally less efficient clock-for-clock than the original Athlon, but the newer P4s are now more efficent clock-for-clock, hence the apparent inequalties in the model rating system.
However, even taking this into account, 3200+ seems overly agressive for that chip, possibly a product of AMD's problems with matching Intel, due of course to the massively late launch of Hammer, or Athlon64.
Hopefully, model numbers will be less of a problem with Athlon64; I think they're planning on moving to a new rating system.
i meant the thing about the cpu speed on the box as if they dont say that 1800+ is 1.53ghz.... in the box my processor came it didnt say 1.53ghz...only 1800+... didnt mean that they were "cheating" hehe...
DWk
It says the the rating and the clock speed on the box.
Digging.Originally posted by DWk@7 September 2003 - 23:21
i meant the thing about the cpu speed on the box as if they dont say that 1800+ is 1.53ghz.... in the box my processor came it didnt say 1.53ghz...only 1800+... didnt mean that they were "cheating" hehe...
DWk
Hole.
Deeper.
Clock speed alone tells you nothing about the speed of a CPU.
An AMD Athlon XP 3000+ runs at only 2.167 GHz, but it kicks the crap out of an Intel P4 running the same clock speed.
Lose your fascination with clock speed, the frequency of the CPU clock is misleading when comparing two different processor architectures.
Okay, one last time, FOR EVERYONE WHO SAYS AMD DOESNT SHOW THEIR CLOCK SPEED ON THE FREAKING BOX.
I took that off my 2500 Box.
Now guys, look, my point in this post is this, and its a small one. (All prices from www.newegg.com, great place to buy parts. and great prices)
AMD 1700+ XP = 60 dollars.
AMD 1800+ XP = 62 dollars.
Now, thats if you go AMD, if you got Intel, here are your options, for SIMILAR AMOUNT OF MONEY.
Intel Celeron 1.7 Ghz = 59 Dollars.
Intel Celeron 1.8 Ghz = 61 Dollars.
Pentium III 1 Ghz = 96 Dollars.
Pentium IV 1.8A Ghz = 127 Dollars.
And if you guys doubt this, here is a link
http://www.newegg.com/app/ListProduct.asp?...rtby=22&order=0
Should work, I think all the queries are saved. And I hope to god none of you Intel fan boys dont try and say, "well there you go, you just proved yourself wrong, that celeron is just as much as the AMD XP." For those of you who are thinking about it, Celerons are the cheap pieces that are in all the really, REALLY low end systems. www.emachines.com
AMD = Best bang for the buck. Like I just did, in your supposed 'real world' situations, dont compare performance, COMPARE THE FREAKING PRICE. I dont know about you guys, but I things get pretty simple in deciding on my proessor if I DONT HAVE THE MONEY FOR ONE OF THE OPTIONS.
I mean honestly guys, first you say, oh well, the 1.8Ghz P4 out performs the 1800+ XP. But, FOR THE 127 DOLLARS, YOU COULD BE AN AMD 2600+ XP. But of course then, you guys try and say "oh, well, you have to pay 80 dollars more than an 1800+. HORRIBLE LOGIC!
well, even then, if the P4 stops at 4ghz or 4.5ghz, that would mean that the final P4s ended up being 2ghz+ faster than the very first P4s. i would call that a successful design, if they're able to get so much mileage out of it. and the same goes for the Athlon, imho.Originally posted by DWk@7 September 2003 - 05:30
i dont think p4 is gonna go higher than 4 or 4.5 since they are putting out these new processors...
Bookmarks