Sign up with Google
Rouxenator
Dank meme lord
- Joined
- Oct 31, 2007
- Messages
- 44,050
I am often accused of trolling intel threads. While trolling actually does not exist, it just means people can't argue a point, let me explain the logic behind why intel really is not a good chip maker to support.
-----------------
Starting out as a memory maker with little success in the processing market intel has always been a company full of hot air. Coincidentally their name would have been "More Noise" if you they had gone with their original plans of using the last name of the two founders Gordon Moore and Robert Noyce (Moore Noyce).
Was it not for IBM trying to get its newly found Person Computer division off the ground, and intel marketing division, we probably would not know intel today. It should be noted that intel’s marketing division is its best performing division after its legal department that never really wins cases but only exists to run up the legal bills of other companies. One could almost say that at intel marketing and legal are the most important divisions behind their success.
So intel got the deal with IBM but part of the standard contracts IBM gave its suppliers was that they need a backup partner so if the supplier goes belly up the second source can support the supply chain. This is where AMD (and Zilog and bunch of others) came on board - they helped intel to secure its deal with IBM.
AMD has had a long history of improving on designs and making micro devices more advance - hence the name. Intel meanwhile was focusing on killing off x86 shortly after giving birth to it (their first attempt at this) with what was know as Intel iAPX 432. On paper it was a powerful 32bit architecture but not having the balls to pull it off intel failed at this by the time their moneycow (x86) was mainstream in its 3rd generation (386).
By this time it also became apparent to intel that they have lost the lead with the 386 and perhaps upcoming 486. Their own 386 topped out at 33Mhz while AMD were making cheap 40mhz chips that could be soldered directly onto motherboards making it an even cheaper alternative to intels socket based CPUs. The 40Mhz AMD 386 processors coupled with a 40Mhz 387 from ST or TI would easily outperform even expensive 486 platforms produced by intel.
At the same time intel found it hard to build its 486 chips without some of them having failures in the integrated math co processor. Being masters at marketing they decided to simply sell these defective chips as a "value line" CPU after lasering off the math co processing unit. Back in the 386 days their marketing department came up with the idea of selling crippled 386s (having only a 16bit external data bus) as the SX range of processors and the full blown 32bit internal and external 386s as the DX range. Now sitting with a bunch of crippled 486s they decided to call the math co less ones SX and the working ones DX. This went so well that later on they produced dedicated lines of 486s that did not include the math co to begin with. Obviously knowing their customers are complete idiots they sold 487 math co processors to them which were in fact working full blown 486DX CPUs with a missing pin. Inserting it into the 487 socket of the motherboard would simple short out the actual 486SX.
Now remember that intel has two departments that it relies heavily upon for success in the industry. Since marketing flexed its muscle by duping buyers into buy dud 486s and then upgrading them to real 486s, it was now time for it legal department to work a little. They took AMD to court and tried to stop them from making x86 based CPUs. They failed (naturally) but did cost AMD a fair bit of time and money in the process. By the time they were done the 4th generation x86 (486) had run its course and it was time for something new. Not being able to trade mark numbers (such as 386/486) they came up with Pentium. The initial Pentium came in two flavours, one running at 60Mhz and one that runs at 66Mhz.
With the initial run of Pentiums intel again showed their fine knack for failing when it had not one but two bugs in it. One being an FPU bug and they had to embark on an exchange program to recall the faulty (aren't they all?) intel pentium CPUs. It should be noted that with the launch of Pentium intel moved to a new socket platform making it impossible to just upgrade to pentium from 486 or 386 platforms. This seems to be a common trend with intel. AMD on the other hand pushed its efforts into providing owners of existing Socket-3 486 platforms with a viable upgrade path.
Intel only offered Socket-3 platform users the insanely over priced Pentium OverDrive CPUs (which was a failure) as upgrade if they wanted anything beyond the 486 DX4-100. AMD went ahead and not only offered the Am486 DX4-120 which outperformed many Pentium systems at the time, but also the Am5x86 range that went up to 133Mhz and could outperform the Pentium 75 from intel. Even more impressive was the Cyrix Cx5x86 which could be over clocked up to 166Mhz. This was a lot of performance on Socket-3 and offered much better value for money than hoping onto Socket-5 with a Pentium CPU that could be outperformed by 486 class CPUs.
On Socket-5 Intel offered the Pentium in speeds from 75mhz to 133mhz. Again they completely missed the plot when they offered OverDrive (their last attempt) in speeds of 125mhz to 166Mhz and OverDrive MMX doing 125Mhz to 200Mhz. Not only were all these options way over priced, they were out performed by the likes of the Cyrix 6x86 and AMD K5. Despite intel putting in a more robust FPU (yeah they learned a lesson with the first Pentiums) the only software at the time that actually bothered with these new FPU functions was Quake. Pretty much all other games and every productivity application of the time used integer calculations. This made the AMD K5 and Cyrix 6x86 look like much better alternatives. The AMD K5 was a marketing failure but technically it was superior to intel's pentium and it had clock speeds up to 133mhz which outperformed 200mhz offerings from intel.
On Socket-7 intel showed their true colours yet again. They backed off Socket-7 soon after it hit the market and did not even stick around to make use of Super Socket-7 boards. What this means is that intel users had to buy a completely new motherboard and CPU (perhaps also RAM) each time they wanted to upgrade. On Socket-7 intel only went up to 233Mhz with the initial MMX range. At the same time they were working on Socket 8 for the Pentium Pro. This was a total failure since they tried to include a lot of level2 cache on the CPU and it was (again) biting off more than they could chew.
Intel was hoping that by leaving socket 7 and not granting anyone else rights to use their new socket8/slot1 design that the competition would be stuck with an outdated platform. Instead AMD introduced Super Socket-7 that supported FSB speeds up to 100Mhz and with the help of VIA that were able to set a quality assurance standard in place which was lacking on previous AMD based motherboards. On Super Socket-7 you could run most old intel socket-7 pentiums going up to 233Mhz, IDT WinChip2 up to 250Mhz, Cyrix MII up to 300Mhz (PR433) and the AMD K6, K6-II and K6-III CPUs that went up to 550Mhz. The K6-2 was remarkable because not only did it offer socket 7 users with a cheap upgrade path but it also included MMX and AMD's own 3DNow! instruction sets. Using the 100Mhz FSB on super socket-7 board also meant that it out performed most Pentium2 CPUs at the same speed.
Speaking of Pentium2, this was one of the few intel success stories, apart from the fact that they locked out competitors from the Slot 1 platform. Pentium2 was actually just a slot version of the Pentium Pro - which enabled them to include the L2 cache chips on the cartridge instead of on the die. They could also now run the L2 cache at half speed which meant a lot less faulty CPUs rolling off the production line. This also helped them to produce crippled (remember the SX range?) budget CPUs more easily and sell them as Celerons. Some of which with no L2 cache at all.
Pentium Pro/II/III used the P6 architecture and this was another success story for intel. They were able to dig it up for Core2 after failing with Pentium 4. Another success intel had on slot1 was the 440BX chipset. AMD usually had to rely on VIA if they wanted a quality chipset for their platforms because chipsets was something AMD sucked at. Intel gave the world the 440BX and it allowed, for the first time, that people could upgrade a little more without having to buy a new motherboard or RAM. Using Slocket adaptors (Slot1 to Socket 370) you could run newer Socket370 CPUs on your older Slot1 board. For once it seems that intel was doing something right. But were they ?
Meanwhile at camp AMD they decided to go with the flavour of the day and they made Slot-A which was one of the few times they copied a trend that was so intel. AMD were so blatant, they got Slot A by just flipping Slot-1 180 degrees. Slot-A was however electrically incompatible (actually superior) to Slot-1. It used the EV6 bus protocol developed by DEC and this was a fairly scalable bus that supported DDR. On Slot-A AMD managed to win the Gigahertz race with little problems. Their K7 range of processors named Athlon overclocked nicely and with this scalability the easily topped 1Ghz.
Intel being desperate to catch up released the Pentium3 1Ghz and it overheated and fried so badly they were stuck with fried egg on their face - nothing new there. They recalled the faulty Pentium3 CPUs (nothing new again) and set about focusing on a new scam with RAMBUS and an architecture called Netburst. The product of this was the Pentium4 and it was a failure right off the mark. AMD Durons (crippled value line) where able to beat Pentium4 CPUs at the same clock speed and Athlons did so at much lower clock speeds.
---to be continued---
-----------------
Starting out as a memory maker with little success in the processing market intel has always been a company full of hot air. Coincidentally their name would have been "More Noise" if you they had gone with their original plans of using the last name of the two founders Gordon Moore and Robert Noyce (Moore Noyce).
Was it not for IBM trying to get its newly found Person Computer division off the ground, and intel marketing division, we probably would not know intel today. It should be noted that intel’s marketing division is its best performing division after its legal department that never really wins cases but only exists to run up the legal bills of other companies. One could almost say that at intel marketing and legal are the most important divisions behind their success.
So intel got the deal with IBM but part of the standard contracts IBM gave its suppliers was that they need a backup partner so if the supplier goes belly up the second source can support the supply chain. This is where AMD (and Zilog and bunch of others) came on board - they helped intel to secure its deal with IBM.
AMD has had a long history of improving on designs and making micro devices more advance - hence the name. Intel meanwhile was focusing on killing off x86 shortly after giving birth to it (their first attempt at this) with what was know as Intel iAPX 432. On paper it was a powerful 32bit architecture but not having the balls to pull it off intel failed at this by the time their moneycow (x86) was mainstream in its 3rd generation (386).
By this time it also became apparent to intel that they have lost the lead with the 386 and perhaps upcoming 486. Their own 386 topped out at 33Mhz while AMD were making cheap 40mhz chips that could be soldered directly onto motherboards making it an even cheaper alternative to intels socket based CPUs. The 40Mhz AMD 386 processors coupled with a 40Mhz 387 from ST or TI would easily outperform even expensive 486 platforms produced by intel.
At the same time intel found it hard to build its 486 chips without some of them having failures in the integrated math co processor. Being masters at marketing they decided to simply sell these defective chips as a "value line" CPU after lasering off the math co processing unit. Back in the 386 days their marketing department came up with the idea of selling crippled 386s (having only a 16bit external data bus) as the SX range of processors and the full blown 32bit internal and external 386s as the DX range. Now sitting with a bunch of crippled 486s they decided to call the math co less ones SX and the working ones DX. This went so well that later on they produced dedicated lines of 486s that did not include the math co to begin with. Obviously knowing their customers are complete idiots they sold 487 math co processors to them which were in fact working full blown 486DX CPUs with a missing pin. Inserting it into the 487 socket of the motherboard would simple short out the actual 486SX.
Now remember that intel has two departments that it relies heavily upon for success in the industry. Since marketing flexed its muscle by duping buyers into buy dud 486s and then upgrading them to real 486s, it was now time for it legal department to work a little. They took AMD to court and tried to stop them from making x86 based CPUs. They failed (naturally) but did cost AMD a fair bit of time and money in the process. By the time they were done the 4th generation x86 (486) had run its course and it was time for something new. Not being able to trade mark numbers (such as 386/486) they came up with Pentium. The initial Pentium came in two flavours, one running at 60Mhz and one that runs at 66Mhz.
With the initial run of Pentiums intel again showed their fine knack for failing when it had not one but two bugs in it. One being an FPU bug and they had to embark on an exchange program to recall the faulty (aren't they all?) intel pentium CPUs. It should be noted that with the launch of Pentium intel moved to a new socket platform making it impossible to just upgrade to pentium from 486 or 386 platforms. This seems to be a common trend with intel. AMD on the other hand pushed its efforts into providing owners of existing Socket-3 486 platforms with a viable upgrade path.
Intel only offered Socket-3 platform users the insanely over priced Pentium OverDrive CPUs (which was a failure) as upgrade if they wanted anything beyond the 486 DX4-100. AMD went ahead and not only offered the Am486 DX4-120 which outperformed many Pentium systems at the time, but also the Am5x86 range that went up to 133Mhz and could outperform the Pentium 75 from intel. Even more impressive was the Cyrix Cx5x86 which could be over clocked up to 166Mhz. This was a lot of performance on Socket-3 and offered much better value for money than hoping onto Socket-5 with a Pentium CPU that could be outperformed by 486 class CPUs.
On Socket-5 Intel offered the Pentium in speeds from 75mhz to 133mhz. Again they completely missed the plot when they offered OverDrive (their last attempt) in speeds of 125mhz to 166Mhz and OverDrive MMX doing 125Mhz to 200Mhz. Not only were all these options way over priced, they were out performed by the likes of the Cyrix 6x86 and AMD K5. Despite intel putting in a more robust FPU (yeah they learned a lesson with the first Pentiums) the only software at the time that actually bothered with these new FPU functions was Quake. Pretty much all other games and every productivity application of the time used integer calculations. This made the AMD K5 and Cyrix 6x86 look like much better alternatives. The AMD K5 was a marketing failure but technically it was superior to intel's pentium and it had clock speeds up to 133mhz which outperformed 200mhz offerings from intel.
On Socket-7 intel showed their true colours yet again. They backed off Socket-7 soon after it hit the market and did not even stick around to make use of Super Socket-7 boards. What this means is that intel users had to buy a completely new motherboard and CPU (perhaps also RAM) each time they wanted to upgrade. On Socket-7 intel only went up to 233Mhz with the initial MMX range. At the same time they were working on Socket 8 for the Pentium Pro. This was a total failure since they tried to include a lot of level2 cache on the CPU and it was (again) biting off more than they could chew.
Intel was hoping that by leaving socket 7 and not granting anyone else rights to use their new socket8/slot1 design that the competition would be stuck with an outdated platform. Instead AMD introduced Super Socket-7 that supported FSB speeds up to 100Mhz and with the help of VIA that were able to set a quality assurance standard in place which was lacking on previous AMD based motherboards. On Super Socket-7 you could run most old intel socket-7 pentiums going up to 233Mhz, IDT WinChip2 up to 250Mhz, Cyrix MII up to 300Mhz (PR433) and the AMD K6, K6-II and K6-III CPUs that went up to 550Mhz. The K6-2 was remarkable because not only did it offer socket 7 users with a cheap upgrade path but it also included MMX and AMD's own 3DNow! instruction sets. Using the 100Mhz FSB on super socket-7 board also meant that it out performed most Pentium2 CPUs at the same speed.
Speaking of Pentium2, this was one of the few intel success stories, apart from the fact that they locked out competitors from the Slot 1 platform. Pentium2 was actually just a slot version of the Pentium Pro - which enabled them to include the L2 cache chips on the cartridge instead of on the die. They could also now run the L2 cache at half speed which meant a lot less faulty CPUs rolling off the production line. This also helped them to produce crippled (remember the SX range?) budget CPUs more easily and sell them as Celerons. Some of which with no L2 cache at all.
Pentium Pro/II/III used the P6 architecture and this was another success story for intel. They were able to dig it up for Core2 after failing with Pentium 4. Another success intel had on slot1 was the 440BX chipset. AMD usually had to rely on VIA if they wanted a quality chipset for their platforms because chipsets was something AMD sucked at. Intel gave the world the 440BX and it allowed, for the first time, that people could upgrade a little more without having to buy a new motherboard or RAM. Using Slocket adaptors (Slot1 to Socket 370) you could run newer Socket370 CPUs on your older Slot1 board. For once it seems that intel was doing something right. But were they ?
Meanwhile at camp AMD they decided to go with the flavour of the day and they made Slot-A which was one of the few times they copied a trend that was so intel. AMD were so blatant, they got Slot A by just flipping Slot-1 180 degrees. Slot-A was however electrically incompatible (actually superior) to Slot-1. It used the EV6 bus protocol developed by DEC and this was a fairly scalable bus that supported DDR. On Slot-A AMD managed to win the Gigahertz race with little problems. Their K7 range of processors named Athlon overclocked nicely and with this scalability the easily topped 1Ghz.
Intel being desperate to catch up released the Pentium3 1Ghz and it overheated and fried so badly they were stuck with fried egg on their face - nothing new there. They recalled the faulty Pentium3 CPUs (nothing new again) and set about focusing on a new scam with RAMBUS and an architecture called Netburst. The product of this was the Pentium4 and it was a failure right off the mark. AMD Durons (crippled value line) where able to beat Pentium4 CPUs at the same clock speed and Athlons did so at much lower clock speeds.
---to be continued---
Last edited:
.