AMD processors

[FynexFox]

Hey, everyone. I have decided to go to AMD processors. I am going to sell my current mobo and processor wich havent even been used to fund my new mobo and processor. Now i know a lot about Intel(unforutnaletly) and not a lot about AMD. Can anyone tell me whats the ebst socket type(939 etc etc) and what like fx and 64 bit and all mean. Alsoi noticed AMD's are very much higher priced. My brother got my intel pentium 4 lga 775 2.66 for 120, i found this amd 754 Operating Frequency: 2.0GHz is like 140, can anyone tell me why? thansk a lot!

[Overconfidence]

1.) Socket 939 is the better socket for AMD, socket 754 is slightly outdated.

2.) An AMD chip at 2 ghz is comparable to an Intel chip at 3.2 ghz. The AMD chip will have a rating of 3200+, which is equivalent to 3.2 ghz Intel.

[Frosty]

veince core is the best core for socket 939 chips the san diego core is the same as the vience just twice the amount of L2 cahce

[FynexFox]

Ahh I see, thats cool. so 1.8ghz is the equiv to 3.0 ghz on an intel? Also whats the series(athlon 64 etc) mean and whats L1 and L2 chache?

[abnoea]

64 is just 64 bit as oppossed to 32 bit, if you go with 64 bit you need to get an appropriate os and mobo.... as for the chache im not too sure what exactly it does/benefits but im thinking its similar to RAM(?)

[Frosty]

the more cache the better the vience core is the best and so is the san diego same core just twice the L2 cahce (san deigo) they run very cool and allow for more Overclocking

[FynexFox]

ahhhh, I see. Sorry for a delayed response, this helps a lot guys! I was going through the 2.0 AMDs and Im looking and I go - 300 bucks! what the crap! This helps a ton guys. Danke*bow*

[Zuke]

Keep in mind that the more cache you get the better it is, but if you get a dual core CPU, then you actually have half the cache you think. Each core will take half the cace for itself. This means that a CPU that has 1Mb of L2 cach actually only is using 512 per core, and the L1 cache is being used faster, forcing the core to goto L2 sooner.

Not that it matters, since the Athlon64 and Athlon64 X2 both have onboard memory controllers to help you goto your DDR faster. Cnet just rated all Dual Core CPUs against each other: AMD swept the field. Enjoy your Athlon!

[Guttenaffe]

Keep in mind that the more cache you get the better it is, but if you get a dual core CPU, then you actually have half the cache you think. Each core will take half the cace for itself. This means that a CPU that has 1Mb of L2 cach actually only is using 512 per core, and the L1 cache is being used faster, forcing the core to goto L2 sooner.

Not that it matters, since the Athlon64 and Athlon64 X2 both have onboard memory controllers to help you goto your DDR faster. Cnet just rated all Dual Core CPUs against each other: AMD swept the field. Enjoy your Athlon!

Cache don't work that way, it really depends on the architecture of the cache system. To put it into simple and quick way to explain it: Cache takes "blocks" of system RAM. For simplicity sake 4 addresses or "rows" of RAM into one line, adds on a decoding TAG which word allocation info and a methood of telling which ram addresses are in that block. Then when you need to execute something it will look in the cache and if the TAG has info that saying that the needed info is loaded into the cache then you have what they call a "hit" or the proc can load the opcode, address, and data VERY quickly and then process the information. If the data isn't there then you have a "miss" and it needs to then be loaded into the cache. While this is happening the processor will go onto the next task and the PC (program counter) will pick it up on the next free/open tick or so.

The problem of cache is that you can only have so much till it slows down the processor. The very nature of the algorithms used in its "hunting" limit the amount you can use in order to have tip-top speed. Don't worry though AMD and Intel know what they are doing and you won't buy a processor with too much cache which will make it slow. So in a way what you said was right, but was wrong. It was right because they will not sell something with to much cache. So the more is the better for now, till one of them starts this marketing scheme that "my proc has more cache than your proc buy mine!" that happened with the clock speeds of processors.

I still say they should measure procs by their "flops" and "mips" written on the box. The only problem is that people would not be inpresed by huge padded numbers.

If you wish I'll give a more detailed example. (this post is long enough)

[Omega]

uh, for socket 754, the Venice ain't the best. it's only got 2.0 core, 11x multiplier, and if i am not mistaken 512kb L2. you want the newcastle. 2.2GHz core, 11x multiplier, 1mb L2 cache. plus it handles OC'ing rather well. i'm running a S754 Newcastle Athlon 64 3200+ OC'd to 2.3GHz (and aiming for 2.5). =/


Keep in mind that the more cache you get the better it is, but if you get a dual core CPU, then you actually have half the cache you think. Each core will take half the cace for itself. This means that a CPU that has 1Mb of L2 cach actually only is using 512 per core, and the L1 cache is being used faster, forcing the core to goto L2 sooner.

wrong. if you find a X2 that claims to have 1mb of L2, it will specifiy that it is 1mb of L2 PER CORE.


64 is just 64 bit as oppossed to 32 bit, if you go with 64 bit you need to get an appropriate os and mobo.... as for the chache im not too sure what exactly it does/benefits but im thinking its similar to RAM(?)

you don't need a different OS if you have a 64 bit proc. however, if you wish to run 64 bit apps (which there are few), then you will need Windows XP x64 edition. about the mobo: it's not like 64 bit mobos are special and exotic either. the majority of AMD boards are 64 bit compatable, for the reason that almost everything AMD makes these days is 64 bit.

[bigbaddog]

The new March issue of "Computer Shopper" has a great article on the new dual core pentiums and amd cpus. Check it out.

[DaveW]

I still say they should measure procs by their "flops" and "mips" written on the box. The only problem is that people would not be inpresed by huge padded numbers.


Mips is an outdated system. Mips, Millions of instructions per second, only measures how fast a processor can decode and execute instructions. FLOPS is a much more accurate measure of the processing power of the CPU. I'm too busy to explain the full situation, but nice description, Guttenaffe. Probably a lot better than i could have done :)

-Dave

[bartvandenberg]

i guess this is kinda the place since everyone here seems to know quite a bit about amd cpu's. ive got a clawhammer 3200+ 64. can anyone tell me how this compares to a venice or newcastle. never seem to here much about them.....