Here is the INFO.

This highway or BUS is an internal connection. On the motherboard level, the HyperTransport bus connects all parts of the motherboard, such as the PCI slots, AGP slots and USB ports to the CPU and memory and also provides the connection between the CPU and memory itself (although it is a bit more complicated than this. The Hammer processors actually connect to memory directly but use HyperTransport to communicate with the memory of other processors in a multiprocessor configuration since HyperTransport is also a coherent bus). At present, the internal PCI bus used in current systems is a rather small road, limited to a small amount of traffic, supporting a maximum data rate of 133 Megabytes per second. This may seem like a lot of data, but when compared to the memory and CPU shoving data around at 2.1 Gigabytes per second and video cards nearing 20 Gigabytes per second, the internal bus is very limiting.

The whole point of HyperTransport technology for the PC is to provide the ability for data to be moved around faster and in larger quantities on the motherboard within computer system.

And here is the link. It's a good read.
http://www.short-media.com/review.php?r=206&p=2

O.K. Here is my understanding about HTT.

With AMD, the CPU communicates with separate HTT links to the system. Independent point to point HTT links are more efficient than Intel’s FSB which everything needs to go through the One FSB for processing.

The PCI bus already enumerates seperate virtual 'channels' for every device on one physical bus, thus the reason for an IRQ table.

The AGP(and now PCIe 16x) bus is routed through its own bus with dedicated bandwidth.

What PCIe accomplishes is having multiple virtual buses through one physical bus on multiple virtual channels, which is how PCIe 1x and 4x devices can operate on the same bus, yet run at different frequencies.

If you look at either Nvidia's Nforce or Via's KT8 series, you still see that there are PCI and AGP buses prevalent, along with an IDE controller and all other peripheral controllers.


Technically, a lot of work would still need to be done to eliminate the channels between the actual component and the controller located on the southbridge.

That's part of the design I'm working on. It basically turns the processor into a spider web, with all of the components branching out from the processor, with full bidirectional simultaneous pipelines to all devices at once.

Well, good luck with that design. It sounds very promising.

I read on the AMD website that All Motherboards made for the Athlon 64 have been specially modified w/ HTT links built in to connect the AGP/PCIe16 & other components.

hmmm would be interesting to know if such design can factor out the i/o bottleneck for IDE controllers.

ramdrives would work perfectly with such design, but they're volatile, so it's not that practical...


ugh, i don't know that much about this stuff... but good luck with it SeanB and keep us informed how it goes

What makes the Bottleneck for the EIDE & SATA drives is there bus speed of only 33, 66, 11, 133, 150 & 300 which is SATA2. The speed which is uses to communicate to the CPU.

AMD's HTT can deliver quick speed but once it reaches the hard drive it lags. Though Intel CPU's lag far greater.

We should have had 500+MHz Hard Drive Bus speeds with 10,000 to 15,000 RPM, 16MB to 24MB buffer cache & with numerous controller arms for quicker memory search.

Hard drive companies are lazy & not keeping up with the industry.

Someone should by out a small non-competitive Hard Drive company & put Western Digital, Maxtor, Hitachi, Sygate etc. to shame with a super design & with the element of surprise.

Super, the external bandwidth surpasses the internal bandwidth... so no, that's NOT the problem.

UDMA133 goes above and beyond what the internal bandwidth is capable of.


Furthermore, AMD's HTT does NOTHING for the IDE/SATA controllers, which are resident in the same place as on Intel chipsets.

There is probably a slight difference in latency because of being able to pass through one fewer controller and channel, but it's a few nanoseconds at most.

Hard drive access times and bus latencies are rated in miliseconds, not nanoseconds, so HTT really doesn't help performance.


The only way hd performance will ever get any better is if they switch to fiber optic sensitive storage mediums through fiber optic channels... either that or holographic arrays.

The Fiber Optic's make sense, but I have to disagree about the HTT. It does prove it's performance over & over again.

This is not my opinion, if you ask me.
It has been proven. It may not be 100% super, but it sure beats the FSB.

In theory, it should. But technically, it doesn't.

All it does is cut one channel and one controller out of the mix.

The southbridge is still architecturally the same, including operating at the same frequency, which is why it doesn't help with any component I/O.

Well, you see Athlon 64's with half the clock speed of an Intel CPU's beating them in performance.
This is a common fact. Even Intel admits to it.

Hi All,

Have been reading your discussions on HT with interest, and thought the following info on nVidia's forthcoming Opteron chipset may be relavent... (as it illustrates a potential use for all that bandwidth.)

(Quoted from: http://www.theinquirer.net/?article=20913)

"Opterons, however, support more than one HyperTransport link - 2-way Opterons, for example, support three. One connects the two chips together, meaning each chip has two spare. Each chip can use one HTT link to connect to the 2200 MCP. The 2200 supports 22 lanes of PCI Express, SATA and PATA RAID, integrated security features and native Gigabit Ethernet. However, with the spare HTT link each chip has, you can also throw in a 2050 MCP and grab yourself another 20 lanes of PCI Express along with more SATA ports. In fact, in a large-scale Opteron system, you can link together a single 2200 with multiple 2050s and get yourself a bandwidth-busting 80 lanes of PCI Express, as well as configuring a RAID array across 16 SATA drives."

Cheers.

I

Sean, quit being so biased. You never bought an AMD Athlon 64 processor. You know how I know?, because it would have proven your biasing wrong with Intel. I can guarantee you that my 3500+'s scores will RAPE your P4's scores like it is now. EVEN OVERCLOCKED! Even with my 6600GT Until you can PROVE that you actually have an AMD and stuff and have tested them, along with having a good gaming pc for yourself, don't flame at anyone or put anyone down over some stupid BS. A processor is a processor. I did my research and I went with AMD, it's as simple as that. Sean, you have to admit, Intel is driving a lot of people away with their heat issues. If you want to have a duel, we can. I can give you my MSN SN and we can swap screenshots of our systems in benchmarks. I want to see what you got. BRING IT ON! I am challenging you if you don't notice.

That's telling him.

I think Sean B is trying to help people understand the PC industry better. But for some reason every word comes out of his mouth is ANTI-AMD & PRO-INTEL with absolutely no factual information to back up what he is saying.

So many of us find facts against what Sean B is trying to feed people & immediately post them, but Sean B’s comeback is that AMD is paying all these review sites to favor AMD over Intel.

So, Sean B, I am kind of curious in to how AMD got so much money to bribe these review sites? Sean B, you are treading in deep water on this one, and I along with many others beg you to heave our warning & stop spreading bulls**t & start presenting factual information. If you are having problems with presenting Facts than don’t post, because you’re close minded & ridicules opinions are not welcomed here at all.

Well said SuperXP. I am not biased, but I know a good processor when I see one, I picked the right choice, the Athlon 64. If there was an Intel that was faster than this w/o all the room-torching heat, I would have gone Intel. AMD seems more promising. Sean, if AMD is sooo unpopular, how come their 2004 Q3 and Q4 earnings skyrocketed because of the Athlon 64? You have got to face it, AMD and Intel is like a wrestling match, one second Intel will be on top punching AMD and the next AMD will be on top mauling Intel (like now).

k so Intel EM64T really 64bit or no if not then can u say that Athlon64 is not 64 bit?

"k so Intel EM64T really 64bit or no if not then can u say that Athlon64 is not 64 bit?"

AMD licensed there AMD64 bit technology to Intel, so they have the same except for little differences.

So, we can't have a PURE 64-Bit CPU, because it wouldn't be compatible. And without compatability right now, people wouldn't buy 64-Bit, because it would require everybody to completely buy new products & such.

Right now 32-Bit & 64-Bit compatability is important.

VERY VERY good points guyz i read all of ur post before i decided to write:

First Sean you seem to know the I/O, north to south however ur problem is ignorance. Because how can u say that AMD does not have a performance again over Intel or the HTT technology which is a better solution than Intels "northbridge" soultion Its beats it out in everyway imaginable and we've seen it time and time again in benchmark after benchmark these sites links, and very bright theads prove that point.

Saying AMD is blowing "hot air" is just plan Bias and is quite frankly just being Ignorant. Please accept this and dont be a fan boy so much when we are just discussing FACTS!

Im not trying to be rude at all but Super and Chris and Ian have all posted valuble data and proof where the AMD "not true" 64-bit HTT (nothing like the Northbridge 800FSB) of 2000ghz CPU beats out the Intel P4 hot as lava soultion which is by the way a fact the pentium 4, 4Ghz CPU couldnt be created just yet because of there heat issue so Intel stopped in their tracks realizing that they were on a clear path off the Yellow brick rode and need a pick up. This is where AMD came in.

AMD decided to help their rivals as they freely share ideas and helped Intel mold their poor arcutechture on the pentuim 4 into a "64-bit" running system although not 64 bit. which we all hear seem to clearly understand. Intel even admits to looking at AMD's arcutechture. So that part of the Bias conversation is bliss.

Also the I/O southbridge does make a difference in performance not only have we seen it. but had Intel Made the HTT CPU intead you would probably have been talking how AMD "800 HT FSB" can't compare to Intels "2000 HTT southbridge soultion" so please do not be ignorant just understand what we are saying. Ive read through every ones post and the "FACTS" are as stated and the "THEORY" is what u speak of (including your spiderman solution)

Sean your better than that and Smarter for that matter dont be an Adam.

Wondefully said tim musomba,

Let's see if this info sticks into Sean B's stubborn head.

Sean, you're a very smart person, I just can't understand why you like to argue about this so strikingly pro-intel like you're living on their payroll.

anyway, back to topic,
this is an interesting article:

http://www.xbitlabs.com/articles/cpu/display/x86-64-rc1.html



benchies in 64bit mode
sandra (uses an intel compiler -biased)
http://www.xbitlabs.com/articles/cpu/display/x86-64-rc1_9.html

independent 64bit software
http://www.xbitlabs.com/articles/cpu/display/x86-64-rc1_10.html

Lost faith,
I've been wondering the same thing about Sean B for a wile now. He can sure give some good knowledgeable advise, but that Intel loving attitude interrupts his help & makes him sound like he is 100% Anti AMD.

Anyway, great links.