Sign in with
Sign up | Sign in
Your question

AMD Athlon 64 3400+ 3.2

Last response: in CPUs
Share
April 13, 2004 12:11:24 AM

In This

http://tomshardware.bizrate.com/marketplace/search/sear...

I think i see an error in it. The error I see is that it says (AMD Athlon 64 3400+ 3.2 GHZ PC, Unix - Socket 940 - Bus 1600 MHZ.) Every thing is correct but it says Socket 940. If im wrong please tell me. I thought it was Socket 754. I checked all other sites about it, and they say its Socket 754. I wanted to let u know about this, and if there is any way i can contact the head people incharge of this site to let them know.

Thanks
Duty

More about : amd athlon 3400

a b à CPUs
April 13, 2004 4:49:22 AM

LOL, everything is wrong. There are no AMD processors that run at 3.2GHz, there are no processors with a 1600MHz bus, the A64 3400+ is a Socket 754, etc etc etc.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
April 13, 2004 5:32:04 AM

Interesting find

-------------------------------------------------
Remember what your fighting for, Remember why you even started fighting, and Remember who you are
Related resources
April 13, 2004 6:05:15 AM

A64 has a 1600mhz HT bus (800 up/ 800 down) @ 16bits.
Though 3.2ghz is wrong. Socket 940 is wrong.

<P ID="edit"><FONT SIZE=-1><EM>Edited by piccoro on 04/13/04 02:05 AM.</EM></FONT></P>
a b à CPUs
April 13, 2004 8:16:59 AM

OMG! That means my P4 with 800 bus also has a 1600MHz bus (800 up/800 down). And my RAM is 1000MHz (DDR500, dual channel). And 2 PIII 1000's really are 2GHz!

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
April 13, 2004 8:32:54 AM

dont you love how advertisers will put numbers together like that? its really false advertising, and pisses me off .. something should be done about it




the A64 doenst have a 1600mhz bus. and its not the same as how DDR @ 200mhz is the same as SDR @ 400mhz

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>
April 13, 2004 12:24:03 PM

>OMG! That means my P4 with 800 bus also has a 1600MHz bus
>(800 up/800 down)

I could be wrong on this, but I think there is at least some merrit to the "1600 Mhz" claim (1600 MT would be more correct), that is that hypertransport is 800 Mhz full duplex, unlike the P4 bus. Over HTT you could send and receive data at 800 Mhz simultaneously, you can not do that on the P4 bus where its either 800 MHz upstream or downstream, but not both.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
a b à CPUs
April 13, 2004 9:58:24 PM

OK, so the speed limit on the freeways here is 140MPH, and my Intel CSA Gigabit controller is really a 2gigabit controller.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
April 13, 2004 10:56:31 PM

>OK, so the speed limit on the freeways here is 140MPH

No, but then following your analogy, intel sells you one way freeways. It sucks when you have to return home driving in the wrong direction ;) 

>nd my Intel CSA Gigabit controller is really a 2gigabit
>controller

If its full duplex, which I presume it is, you could probably claim it is 2 GT/s. Surely you will agree a halfduplex 1 Gb nic isnt as as good as a full duplex one ?

= The views stated herein are my personal views, and not necessarily the views of my wife. =
April 14, 2004 12:05:03 AM

Despite popular opinion, driving in opposing traffic is doable, just that it will involve a lot of honking, swerving, and screaming.

SEX is like math. Add the bed, subtract the clothes, divide the legs, and hope you dont multiply
a b à CPUs
April 14, 2004 1:19:57 AM

The problem I see here is that it will never perform like it has 2x the bandwidth of Intel's 800 bus. It WILL perform like it has less latency, but that's mainly due to the on-die memory controller, not the chipset northbridge.

Intel still calls their Gigabit LAN just that, even though it is full duplex. That can't last long, AMD keeps pushing up the numbers. AMD was so successful calling their 100MHz bus 200MHz that Intel had to follow the same sham, er, scheme when they introduced a QDR bus.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
April 14, 2004 1:28:49 AM

You take numbers to literal.
You should just let it go.
They say 1600mhz because it is 800mhz up and 800mhz down at the same time which means you can actually push data in both directions and it will be the same as a half duplex 1600mhz bus. But it is at 16bit rather than 32bit.
Which you know means it has the same bandwidth as the P4 but just more efficent.
It is marketing but nothing new.

You refer to the P4c's FSB as 800 don't you?
I doubt you say it's a 200mhz bus that is quad pumped.
Or that the Athlon XP's bus is 400 rather than saying it's a 200mhz FSB that is double pumped.
To me, that's the same thing as this.

If you know your stuff, you can go with the flow and not get in a tangle.
E.I.
Sure AMD's performance numbers (especially on the AthlonXP) are out of wack but if you know the CPU behind them why should you freak out? Unless you want to freak out...
Get my point?

<P ID="edit"><FONT SIZE=-1><EM>Edited by piccoro on 04/13/04 09:57 PM.</EM></FONT></P>
April 14, 2004 2:16:49 AM

The problem with those numbers is simple, really: There are at least two degrees of mathematical freedom here which are highly relevant.

One of them is <b>transfers every second.</b> The higher the number of transfers, the lower the latencies will tend to be. The other is <b>bandwidth.</b> Thus saying that a 1600MT/s HT bus is faster than a 800MT/s P4 bus is inaccurate in bandwidth... Both offer 6.4GB/s. Perhaps it would be more adequate to do with busses what they did to memory: PC3200 transfers 3200MB/s. So A64's HT bus is a duplex 3200, and P4 is a 6400 bus.

What I think is confusing is that, right now, some advertise A64 as having 800Mhz HT bus, and some advertise a 1600Mhz bus. This is transfer frequency, but it's a little stupid... It confuses people.

Ideally, everyone ought to understand these things, but they don't. Would a full duplex at 800Mhz be better than a half at 1600Mhz? Absolutely no. The 1600Mhz could allocate the whole 1600 million transfers into one direction, if needed. So advertising the A64 as having 1600Mhz is quite erroneous. It would be better to just really advertise without trying to make your product look good. HT is 800Mhz, full duplex, 16-bit. P4's bus is 800Mhz, Single Duplex, 32-bit. Still not complete, but getting better.

Even alienware is doing it nowadays... In order to just illustrate the mess, I'll pull a little <b>misleading information</b> that is on alienware's site. They mention that HT bus on A64 works at 1600Mhz, and they mention 12.8GB/s maximum bandwidth. But HT only offers 3.2GB/s bidirectionally, which is still not necessarily better than 6.4GB/s unidirectionally. Here's the quote:
Quote:
HyperTransport technology implemented in the AMD Athlon 64 series of processors enables the system bus to operate at 1600MHz. The total delivered processor-to-system bandwidth is up to 12.8GB/sec allowing for high speed I/O communication and increased performance.

Alienware must be getting that 12.8GB/s number from summing dual-channel memory with HT bus speed... Not a fair move in my book of rules. It's a marketing problem; if you look into it, it all feels as if someone is trying to bloat the numbers. They shouldn't sum numbers arbitrarily; this gets confusing really fast. And for those of us who know what we're dealing with, it's really annoying.

In reality, of course, HT doesn't bear the burdain of memory access like P4's bus. So it's different; but it's not to say that AMD's processors have a 1600Mhz "traditional" bus capable of 12.8GB/s as implied by alienware.

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i><P ID="edit"><FONT SIZE=-1><EM>Edited by Mephistopheles on 04/14/04 01:24 AM.</EM></FONT></P>
April 14, 2004 6:46:26 AM

I have no problem whatsoever with simplifying 200 Mhz DDR into 400 MHz or 200 QDR into 800. Using "pure" Mhz and a quantifier/multiplier would lead to too much confusion and would really not tel tell you anything more. This has pretty much always been done, by everyone in the industry, wether it is with FSB's, DDR Ram, GDR RAM, anything.

Full duplex/ half duplex is something else, I agree that calling HTT 1600 Mhz is misleading, though it remains just as true even at 800 MHz, it is better than a half duplex 800 MHz bus, in theory, up to 2x as fast.

>The problem I see here is that it will never perform like
>it has 2x the bandwidth of Intel's 800 bus.

Sure it does; in theoretical/borderline cases it could even have much more than 2x the bandwith. THe P4 has to share its FSB for memory bandwith and I/O, while K8 can use HTT for I/O only, and full duplex. iF you find a way to saturate HTT with AGP/PCI(-E) devices, in both directions, while maxing out memory bandwith as well, you could even see up to 4x the overall performance (bandwith) of AGTL+. You will never achieve this with 66 Mhz PCI, but with PCI-E you could come a lot closer

> That can't last long, AMD keeps pushing up the numbers.
>AMD was so successful calling their 100MHz bus 200MHz that
>Intel had to follow the same sham, er, scheme when they
>introduced a QDR bus.

LOL, yeah blame AMD. First, its oem's advertising the "1600 MHz" bus, not AMD, and also, I'm sure intel never played to the number game, and had a big problem using RDRAM 600/700/800 for 133 MHz modules (which in turn lead to those stupid PC1600/2100/3200/.. naming)

= The views stated herein are my personal views, and not necessarily the views of my wife. =
April 14, 2004 6:52:01 AM

how many times have we seen messing with numbers for marketing perpouses? its ntohign new, while i think its kind of stupid to have those things, it sells and companies do what it takes to sell parts lol. amd is as prone to this as intel. personally, i dont care what marketing either company uses. the normal non informed computer buyer will of courses see bewildering numbers like fsb, ghz, and now two sets of model numbers. i feel sorry for all of them lol . ''ok should i get the amd 64 3200+ or the intel p4 550?'' there is no end to the confusion lol. last decade it was mhz, now its the model or 'pr' number.
a b à CPUs
April 14, 2004 7:52:29 AM

LMAO! D00D, PC600 ran at 300MHz clock, using DDR technology for "600MHz data rate". PC800 ran at 400MHz clock, PC1066 ran at 533MHz clock...these are real clock speeds here, hence so many people were trying to figure out whether their board had 400MHz or 600MHz DRCG's (Dynamic Rambus Clock Generators, I believe they were called) when overclocking and using faster RAM.

<font color=blue>Only a place as big as the internet could be home to a hero as big as Crashman!</font color=blue>
<font color=red>Only a place as big as the internet could be home to an ego as large as Crashman's!</font color=red>
April 14, 2004 8:06:55 AM

Okay, I stand corrected, you're right on this one. Not that it changes my argument in any way though.

= The views stated herein are my personal views, and not necessarily the views of my wife. =
!