The real problem is, it's impossible to decide which of these threads to make the obvious point they're missing ... you can't set market leading prices unless you're a market leader.Originally Posted by Tabbykatze
I mean, I've said it here, but do I go back and post it in the AMD quarterly results thread as well, just in case? Or does that end up with me just spamming the forum repeatedly with the same message, as infinitum....
Message to Intel: If you want to "starting from zero" the first thing you have to do is to replace Raja Koduri and Bob Swann.
How does it go, the definition of insanity is doing the same thing over and over again?The real problem is, it's impossible to decide which of these threads to make the obvious point they're missing ... you can't set market leading prices unless you're a market leader.
I mean, I've said it here, but do I go back and post it in the AMD quarterly results thread as well, just in case? Or does that end up with me just spamming the forum repeatedly with the same message, as infinitum....
The difficulty with making a point is how it is presented, as a nice straight line or as a tangled ball of wool
I for one have optimism that people can learn from past mistakes! Raj isn't a name that excites, but, I don't see AMD suddenly bringing out amazing GPU's now he is gone either, plus he is far more knowledgeable and talented than most and still deserves respect (Intel seem to agree too), so let's see what time tells us.
Re: Intel talks up its discrete GPU plans and aspirations
But he is a happy chappy you must admit, what with all that LOL-ling about
------------------
Valar Morghulis
It didn't work out well for VIA when Intel went after them did it. VIA used to be pretty much everywhere, in the P4 era when Intel were contracted to only make RDRAM chipsets that no-one wanted it could be argued that the DDR chipsets from VIA saved Intel's bacon but made them big enough to become a target.Hmmmm, nvidia to buy up Via for the x86 license?!
Hopefully they can hit it out the park at the very start, inject some realism into GPU prices.
Intel are not exactly known for reasonable pricing!
That is very true, but historically that's maybe because they didn't really have any competition from AMD and they were the kingpins of CPUs. Now they're going into a space with 1 very dominant company at the high end and another company who're starting to make great bang for buck cards from the low end up to upper midrange.
Intel aren't going to be able to build something ok and then charge a premium price for it, they've got to incentivize people leaving AMD/NVIDIA (brand loyalty being a thing). So I reckon their best bet would be to build something towards the upper midrange (GTX 1080/Vega 64 levels) but price it about £320-350. That'd be a big statement but I've no idea if that's possible. Certainly with the size of company, experience of making chips and boards and their own fabrication plants, they're in a better position to drive costs down as much as possible than AMD/NVIDIA who are subject to TSMC/Global Foundrys/Samsung's fab plants.
Its not as simple as throwing money & people at it. The GPU field is a veritable IP minefield where Intel has to tread carefully not to get sued to kingdom come. It will have to develop a lot of its own IP or license it to deliver what it claims it wants to.
OK, let me put it another way...
Intel were traditionally a DRAM company. Many years ago they pulled out of a perfectly profitable DRAM market because the level of profit wasn't high enough for them. The statement at the time was that they only wanted to make silicon with a high profit margin. That has been their way ever since.
The only way that Intel will produce graphics cards at a consumer friendly price will be as part of a strategy to get enough market penetration to then crank the price right up and milk us for every penny they can. If they can hurt Nvidia along the way, then they will consider that a big bonus.
Intel have held back the CPU market for decades with the worst instruction set available in mediocre chips. Take a good look outside the PC market, Intel are only successful in the PC form factor. Their masterstroke was getting companies to decide that the PC was good enough for server use. If you want a stagnant GPU market as well, then they should get your support.
Yes, far too many people seem to have forgotten or never knew how truly awful x86 really was. Especially that the original i8086 was chosen by IBM for the original IBM PC versus for example the Motorola 68000K held back computing for years. 64KB segmented, far to few registers, 1MB max address space, etc. What a truly awful architecture Intel came up with.
You mean AMD's masterstroke that ended up helping IntelCough AMD64 cough. Their Itanium didn't exactly do well
"In a perfect world... spammers would get caught, go to jail, and share a cell with many men who have enlarged their penises, taken Viagra and are looking for a new relationship."
You sound like someone who had to deal with the likes of EMS memory and himem.sys
It wasn't just that though, Intel's mastery of politics and a bullying business style in the 486 era where they promised the Pentium would be the fastest thing ever and other companies should give up on processor design was sadly successful enough to kill off the Clipper CPU and give Silicon Graphics a few laps around the drain hole.
Oh it started well before then, I remember people wondering if they should get a SPARC server which they actually wanted but ended up with a 486 or Pentium because it could scrape by for half the money.You mean AMD's masterstroke that ended up helping Intel
Itanium is an oddity, seeing how AIUI it was basically a Hewlett Packard designed chip that Intel managed to get the rights to and try to stuff down our throats regardless of how much we might be gagging on it. If Intel were going to push a non x86 chip for the 64 bit transition, they could have at least made it the DEC Alpha which I believe they still have the rights to.
But basically I am struggling to think of anything Intel has produced that was both genuinely new and good. Pentium was a static dual issue design where most of the "innovation" like register scoreboarding had been done by Seymour Cray decades before. Compare that to the AMD K5 which although it was a more modern OoO design sadly flopped because it had a low latency FPU optimised for spreadsheets etc and not a pipelined FPU that was used in Quake. But a MIPS chip could wipe the floor with either of them.
Itanium concept was new, as was the i432 and the concept of an FPU stack rather than registers. The double clocked P4 core was original as well. But those all sucked. Oh, and MMX. Who the heck wanted integer only vector.
Last edited by DanceswithUnix; 07-02-2019 at 08:56 AM.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
LinkBack URL
About LinkBacks
Reply With Quote