400fsb...SO WHAT!!!!!

[AfroGeek1]

OK ...The norm now is a 333FSB. If I remember correctly most of the decent 333 motherboards can get up to 400 with no problem. Please correct me if I'm wrong. My point is what's the point? Give me a meal, not a snack! You know what I mean ? Give me an athlon that will do 800fsb ! Shoot, sounds like to me they need to work on the CPUs not the motherboards.

 

[vk2amv]

800FSB is just an intel myth. It is only 200Mhz. Same as with the Athlon 333 is a myth. It is 166. The current intel standard it is 133Mhz for the current P4`s and the canterwood chipset will bring that up to 200Mhz. AMD at the moment is a standard of 166Mhz soon to be increased to 200Mhz. So at the moment AMD have the true faster FSB speeds at 166Mhz and soon they will both be the same at 200Mhz.
AREA_51

'It's only when you look at an ant through a magnifying glass on a sunny day that you realise how often they burst into flames'

 

[TKH]

An Athlon XP on 800FSB? You must be day dreaming. You know how difficult to increase the FSB. I believe you are an Intel fan and want to show off the "800FSB" Intel CPU but please study first before flaming. The 800FSB of Intel is also running at 200MHz and uses QDR multiplier which make the processor a 4x myth. And decent 533FSB Intel chipset mobo can also go upto 667FSB if you overclock from 133MHz to 166MHz, that's nothing but number games. OMG, why do I even bother this?

You never know how stupid you are until you have done something stupid enough for you to realize it.
<A HREF="http://www.anandtech.com/mysystemrig.html?id=22996" target="_new">My System Rig</A>

 

[Mephistopheles]

The data transfer rates on the current P4s is indeed 533Mhz. That´s no myth. The data transfer rates on the AXPs is also 333Mhz. Adressing, however, which is not bus-intensive, still occurs at 133Mhz and 166Mhz, respectively. But it´s data transfer that fills up the bus, so it´s perfectly reasonable to state that these FSB speeds will transfer as much data as a "full", non-multi-data-rate FSB at 533Mhz/800Mhz/333Mhz would transfer. No myth there! And AXPs FSB is not the "truly faster" at all. People should stop trying to find myths so desperately! As if "Mhz Myth" wasn´t enough, we´ve now invented the "FSB Myth"!

 

[vk2amv]

The fact of the matter is that you may get the bandwidth in theory but you also get the latencies from the TRUE slower bus speed. The latency of the P4 bus at 533 fake Mhz is something like 32 last time I looked. Thats around 2.5 at 133mhz. So no you dont get the same transfer as a true 533mhz bus. Its fine though for the P4 because the chip is a high latency CPU by its nature and depends on the high memory bandwidth to try and compensate for the ineffeciencies in the CPU. Which is a system that seems to work for now. The best explination of comparing the P4 and AMD FSB`s is the P4 FSB is like a big truck and the AMD FSB is like a race car. The P4 FSB/truck takes longer to get going and to change directions but it can transfer much larger chunks. The AMD FSB/race car cant transfer as much in one go but it can get going faster and change directions a lot quicker. Now you quoted "But it´s data transfer that fills up the bus, so it´s perfectly reasonable to state that these FSB speeds will transfer as much data as a "full", non-multi-data-rate FSB at 533Mhz/800Mhz/333Mhz would transfer." I will respond to that here. Yes you are correct it would be perfectly reasonable if they quoted the effective speeds. But they go and try and state that these are true speeds. That is what my gripe is. I dont remember who started the lie first but put simply stating that they are true bus speeds is just false advertising. I would say it was intel that started it first because when the P4 came out it was bad because the P4 only had a standard 100Mhz FSB and AMD had the 133Mhz FSB for some time.
AREA_51

'It's only when you look at an ant through a magnifying glass on a sunny day that you realise how often they burst into flames'

 

[Mephistopheles]

would be perfectly reasonable if they quoted the effective speeds. But they go and try and state that these are true speeds.

You know, the one who actually started these things were the <i>memory manufacturers</i>. Or do you really think DDR400 runs at 400Mhz? It doesn´t, it runs at 2x200Mhz, at they still get away with it.

I think your gripe there is only about a valid architectural difference, but I can´t see that as a lie (if it is indeed, everyone´s doing it ever since plain SDRAM got old...). And about advertising, they quite often mention it´s an "effective 400Mhz FSB" in ads... Not often enough, maybe...

But the architectural differences don´t actually allow for the freedom of speech that is saying "AMD´s FSB is faster". That´s technically incorrect - actually, incomplete. I would interpret these numbers - 533Mhz, 400Mhz, 333Mhz, 800Mhz - a little more loosely than you are. As in DDR. They give an idea of how much data can be transfered from Northbridge and CPU, a.k.a. the <i>bandwidth</i>, which is important to consider. So they can, at least in my radar, get away with this. I actually agree with this; it is a marketing scheme which is comparable to AMD´s PR rating. Sure, it omits some information regarding latencies, as you put it, but you must consider that there´s another information regarding the architecture of these FSBs embedded in the bandwidth number you see when you look at the simplified specs... Anyway, I have some old ads from Intel around here, and in every technical specification you actually see 100Mhz-Quad -> effective 400Mhz... So I guess what I´m saying is: OK, so it´s not 800Mhz FSB, but loosen up! It´s not "fake" in that sense, it´s just the bandwidth number. And they never said it was "true"; most of the time, Intel states it´s effective. Sometimes, it doesn´t, but hey, take it easy. No need to go telling everybody "it´s a lie, those *** liars started it"... You might try "AMD has lower latencies inherent to architectural differences in their FSB"... You´d certainly get away with <i>that</i>.

 

[Mephistopheles]

One thing I noticed from your posts, though... You <i>like</i> AMD and <i>hate</i> Intel, and you set your mind on that alone, don´t you? That´s called "being biased". It has become pretty commonplace lately, I´m sorry to say...

You call Hyperthreading a "flop". The way I see it, that´s very naive of you to say. You also state that P4s suck at multitasking, but fail to recognize the usefulness of HT at it. I don´t get that, sorry. You repeatedly state that the P4 is an inefficient CPU but fail to acknowledge the fact that right now, the fastest CPUs on the market are P4s. They have low IPCs and high clock rates, as opposed to AMD, but right now, they´re mostly winning the performance race with that net performance (IPC*clock). In executing SSE2-written code, which is becoming more and more common nowadays, the P4 can´t possibly be called inefficient. And right now, you´re calling Intel the "original sinner" regarding that QDR-FSB of theirs. Take it easy, vk2amv...

In the end, you must avoid thinking that one processor is an absolute winner over another, quite simply. It´s too simplistic to be true, and we must keep our minds open to reason. And to new things, like HT and x86-64.

 

[AfroGeek1]

Thanks for your input. However I see you don't get my point.

1. Do you really think it's hard increasing the front side bus? I don't!! More than likely AMD has already built a chip that does it right this minute. However, it's probably their 2005 model. This is how industry works. More than likey we are all playing with CPUs that they designed back in 2000 or 2001. You're funny, you act like they just made the sh**...lol

2. I understand the math and no i'm not an Intel fan. But, you have to realize that most decent AMD motherboards get hella close to a 400mhz (200x2) bus right now. I know 800 may be asking too much for them to release right now. but dayum, don't have me running out to the nearest PC outlet for a measily 33mhz more of performance wich un-mythed is really only a 61.5 MHZ performance boost. Please tell me there something more magical about the newer chipset, which was probably built a few years ago too. If a chipset designer or motherboard company wants to impress me, give me more features at the same price point.

For example: The Nforce chipset is the baddest chipset out there. If you go AMD you go Nforce. BUT!!! In the days of dual channel DDR, what's up with this 3some dimm slot crap?

Oh yeah..let's say we want to keep our ATA133 Raid arrays. There is not one Nforce MB that has an IDE raid controller. I mean real IDE not that SATA half and half crap. Those SATA/IDE adapters cost 30 bucks a piece. Add that to the price of your motherboard! After a while I simply stopped looking for an AMD motherboard with the features I needed and simply looked for "ANY" motherboard with the features I needed.I ended up with an Gigabyte SINXP. It just happened to be an P4 motherboard, which at this point is the only MB I could find that had dual channel ddr, IDE raid, and SATA raid. The hyperthreading don't hurt either. I would have been happy with my barton 2500 if Nforce (wich to me is the only worth while AMD chipset) had a chipset that included the features I was looking for.

4. Another thing. I like the MSI K7N2G deluxe . but dayum! A state od the art motherboard and they're still messin" with 10/100, while everyone else is throwing a gigabit at us. Ok Ok so I may or may not need it. But, let me decide that. Not only this the new 400mhz Motherboard looks suspiciously like the K7N2G deluxe. It just has a kool looking "400FSB" sticker on it. Shi** give me an extra dimm slot! Give me an extra IDE port with an IDE raid chip to boot(preferrably highpoint), get a silicone graphics SATA and drop that weak promise SATA chipset! Why just 2 sata connectors? Give me 3 or 4 so I can do some real raid sh**. Why not have a built in feature that allows users to fully unlock their CPUs instead of using wire,glue,and a whole lotta risk? I know AMD would be pissed, but so what! We're gonna do it anyways!

 

[bikeman]

One remark to Mephisto: Adressing on the P4-bus happens at twice the actual clockrate, i.e. twice per clock. But there are four data transfers per clock, that is correct. Now let's read the rest of this thread ...

Greetz,
Bikeman

<i>Then again, that's just my opinion</i>

 

[bikeman]

Djee, Afrodude ... Chill, man ... You apparently have had a bad day ... Motherboard manufacturers do whatever they want. Some mobo's are loaded with features, other are quite basic. And yes, it is quite hard to find the hardware that exactly fits your needs. But why make such a point of it? You were just yelling at the poor people here, because some mobo manufacturers don't do it the way you want ...

Again: chill, my dear friend, chill ...

Greetz,
Bikeman

<i>Then again, that's just my opinion</i>

 

[Mephistopheles]

Ooops... Sorry there. That wasn´t right indeed. Adressing happens at twice the actual clockrate on the P4-bus. Thanks for pointing that out. It also happens at twice the actual clock rate on the Athlon, as far as I know.

 

[TKH]

Well I'm not flaming at you and I apologies for accusing you as a fanboy. What I meant was jumping from current 166MHz FSB to the value you're asking (400MHz FSB) is impossible on current situation.
I know you are upset by the motherboard manufacturers, but this is the game and the rules. Unless you own your own manufacturing facility or else you can't change anything. I'm glad you can find a motherboard which suit you. Like I always said, think before you buy and build a system to suit yourself by looking at what is your requirement. There's no point yelling at AMD when they are not the one who make mobo. You've ponted your gun at the wrong person.

You never know how stupid you are until you have done something stupid enough for you to realize it.
<A HREF="http://www.anandtech.com/mysystemrig.html?id=22996" target="_new">My System Rig</A>

 

[vk2amv]

Granted I prefer AMD to intel but most of the "crimes" you stste there are not true. Yes I call hyperthreading but I have also stated that it has potential in the future if they can improve the branch prediction to try and stop mispredictions. And I have never once said that at the moment AMD is faster. In fact a few times I have recommended intel over AMD as the 3.06 P4t is untouchable right now by AMD except for overclocked CPU`s. You maybe right about it being the ram companies starting it. I dont really think anyone remembers who started it. And yes intel did start off by saying 100mhz quad pumped but gradualy they started to claim that it was a true 400mzh bus speed. And I call a CPU that needs 2.4gighz to match another CPU at 2.0gighz that is built on a much older design and architecture ineffecient. But the point to all this is infact you are the fanboy most of the time. You are not a real bad fanboy but you still are. Then again so am I. We are pretty much in the same boat I guess. We both support the CPU`s we "grew up on".
AREA_51

'It's only when you look at an ant through a magnifying glass on a sunny day that you realise how often they burst into flames'

 

[Mephistopheles]

Hm... I guess we should both try to keep our minds open.

 

[Crashman]

Actually AMD called their original DDR bus 200MHz far before DDR memory was available for the PC motherboard. Such chipsets as the AMD Irongate, VIA KT133 and KT133A, used SDR SDRAM exclusively.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>

 

[Mephistopheles]

Really? Didn´t know that... I think I didn´t really pay that much attention to PC hardware until recently, so I definitely didn´t know of AMD´s plans (or Intel´s, for that matter) until a very few years ago... Thanks for clearing that out, anyway.