Dispelling the 64-bit/32-bit MHz Myth

[Shrek]

A lot of people seem confused about MHz gaps between IBM's PowerPC 970 chip and Intel and AMD's 64-bit chips. Well, the truth is, there is no gap. Here is proof:

http://www.intel.com/products/server/processors/server/itanium2/index.htm?iid=ipp_browse+featureprocess_itanium2&

The Intel® Itanium® 2 processor is uniquely architected for demanding enterprise and technical applications. Itanium® 2-based platforms enable businesses and organizations to maximize their investments by delivering industry leading performance at lower cost with greater choice than proprietary technologies.

Available Speeds: 1 GHz, 900 MHz
Cache Level 3: integrated 3 MB or 1.5 MB
Level 2: 256 KB
Level 1: 32 KB
Features
Based on EPIC architecture
Enhanced Machine Check Architecture (MCA) with extensive Error Correcting Code (ECC)
Operating system support: HP-UX*, Linux*, Windows*
System Bus 400 MHz, 128-bit wide
6.4 GB/s bandwidth
Chipset Intel® E8870 chipset, OEM custom chipsets

I could not find information on the clock speeds of AMD's future Hammer processors, but if anyone else can please post them here.

Bottom line is, there is difference in the overall performance of 64-bit chips vs. 32-bit chips; 64-bit are faster in overall performance than 32-bit chips.

Hope this helps. If anyone on MacRumors can better explain things, please do. Thanks.

 

[springscansing]

I don't know what this is supposed to be proving. People are concerned about the gap between the 970 and the P4. The P4 as it is right now kills an Itanium 2 in most things.

 

[Catfish_Man]

Hah!...

Originally posted by springscansing
I don't know what this is supposed to be proving. People are concerned about the gap between the 970 and the P4. The P4 as it is right now kills an Itanium 2 in most things.

...the Itanium2 and POWER4 are roughly tied right now for fastest processor in the world. The P4 can beat the Itanium2's integer score, but it gets squished at floating point. Also, it can't do multi-processor systems (unless you count the much more expensive Xeon).


On topic:
The reason the Itanium is so much faster, at such a low clock frequency, has NOTHING to do with it being 64 bit. It's based on a completely different design philosophy than either PowerPC or x86 chips (PowerPC being RISC, x86 being modified CISC) called VLIW. This philosophy shifts almost all of the work of scheduling instructions and figuring out how to execute more than one instruction at once from the chip to the compiler. Theoretically this should result in small, ultra-high performance chips that require extremely sophisticated compilers (and have long compile times). In practice, the Itanium2 is fast through brute force. It's HUGE (mostly because of an enormous on chip cache, and an entire x86 core for 32 bit code), expensive, and power hungry. Yes, it's fast, but it's not that fast for its size/price. The PowerPC 970 is not in the same market as the Itanium (or its big brother, the POWER4). Its main competitors are going to be the Athlon64 (ClawHammer) in 1 and 2 processor systems, and the Opteron (SledgeHammer) in 4-8 way systems. From initial SPEC numbers it appears that the PowerPC 970 will be slightly lower performing (especially in integer code), but use much less power, have a lower pin count, a faster memory bus, and be smaller. The low power version of the 970 should be very suitable for laptops and 1U servers, which is likely going to be difficult for Hammers. Overall, a 1 processor 970 system will be competetive (and win handily at vector code), and a dual 970 system (or, even better, dual core) will be well ahead of the competition.