fy/joe/jc thx for the replies
Is it now an accepted urban legend that SSE2 will be implemented in the Hammers? My recall is that floating point enhanced instructions will be implemented, not the entire instruction set.
fyodor re:P.S. Any "perpetual license for x86 instructions" would not cover MMX, SSE and SSE2.
Did that "grandfather" pass away with the 486?
---------------------------------------------------------
To:dhellman who wrote (28411)
From: fyodor_ Wednesday, Feb 14, 2001 7:43 AM
View Replies (1) | Respond to of 28507
dhellman: Would AMD trade HyperTransport for SSE2?
AMD cannot trade LDT for SSE2 or anything else. It's a free, (relatively) open standard.
Ultimately, will Intel have to use HT-Technology to be competitive?
Intel would likely be willing to go a long way to avoid it, but only time will tell. If it becomes a universal standard like PCI, sure they will have to use - and they will (in that case).
LDT is, imho, all about influence for AMD. All the publicity probably won't do much harm to the stock either ;)
-fyo
To:fyodor_ who wrote (28414)
From: John Petzinger Wednesday, Feb 14, 2001 1:09 PM
Respond to of 28507
fyo, "AMD cannot trade LDT for SSE2 or anything else. It's a free, (relatively) open standard. "
I'm not a lawyer, but I believe AMD would prohibit Intel from using the HT trademark in any of their advertising. Also, if Intel is not a member of the consortium which will eventually form to mediate changes/improvements to HT, Intel will always be one step behind.
Its a similar situation with SSE2 - Intel cannot prevent AMD from using the instruction set, they can only prevent AMD from calling it SSE or SSE2.
Petz
To:dhellman who wrote (28411)
From: Jozef Halada Wednesday, Feb 14, 2001 12:12 PM
View Replies (1) | Respond to of 28507
dhellman,
Would AMD trade HyperTransport for SSE2?
I think AMD said that it is not going to use LDT to barter for other patents, and my guess is, Intel is free to become a licensee.
As far as SSE(2), I am not sure it is patentable. I seem to recall that an instruction set cannot be patented, only implementation (correct me if I am wrong), and I am pretty sure AMD will implement it differently. Besides, I think AMD has a perpetual license for x86 instructions
Joe
To:Jozef Halada who wrote (28447)
From: fyodor_ Wednesday, Feb 14, 2001 12:23 PM
Respond to of 28507
Joe: As far as SSE(2), I am not sure it is patentable. I seem to recall that an instruction set cannot be patented, only implementation (correct me if I am wrong), and I am pretty sure AMD will implement it differently. Besides, I think AMD has a perpetual license for x86 instructions
Correct, instruction sets are not patentable. Intel found that out with MMX. The question remains how easy it will be to implement SSE2 without violating any Intel IP. People more knowledgeable than I have raised concerns over this, in any event.
P.S. Any "perpetual license for x86 instructions" would not cover MMX, SSE and SSE2.
-fyo
To:dhellman who wrote (28411)
From: jcholewa Wednesday, Feb 14, 2001 1:21 PM
Respond to of 28507
> Should AMD trade licensing of HT-Technology for free inclusion of the SSE2 instructions
Intel can't patent the instructions. AMD can use them pretty much freely as long as they don't use the same exact implementation. Or so I recall.
-JC
------------------------------------
and from the Bull:
By: sgolds $$$$
Reply To: 65738 by dhellman $$$$ Wednesday, 14 Feb 2001 at 8:19 AM EST
Post # of 65795
Dhellman, AMD already has the rights to SSE2 (it will be in Hammer) as part of a settlement with Intel some years ago.
What's the relative value of a bus standard vs an instruction set standard?
A good 'bus' standard is way more valuable than an instruction set. A very good argument can be made that Intel rose to prominence in the industry on the development of the PCI bus. Before that, Intel was one of several leading chip makers in a competitive industry. With the developement of PCI, Intel wrested the control of microprocessor developement away from IBM and surged into hte industry lead.
Note how many companies have lined up to do technology swaps for the rights to use HypeTransport. That is value!
Technical semantic point: HyperTransport is not a bus, it is a fabric. That is the beauty of it: Multiple devices can talk across the fabric at full speed. On a bus, every device shares the bandwidth with every other active device. On a fabric, everyone runs a full throttle. |
