Dubious Distinction

[unhuman]

https://www.extremetech.com/computing/274650-the-worst-cpus-ever-made

[JB]

Haters gonna hate.

[sparkdrummer]

TI’ers are like bees. They don’t know they shouldn’t be able to fly. We don’t know that our processor is so crippled and crappy so we make it do wonderful things. 

[+jedimatt42]

I wish there was an add blocker that blocked these micro-blog pages entirely. 

[+OLD CS1]

5 minutes ago, jedimatt42 said:

I wish there was an add blocker that blocked these micro-blog pages entirely. 

Easy:

 

/etc/hosts

 

127.0.0.1 www.extremetech.com

[+acadiel]

4 hours ago, unhuman said:

https://www.extremetech.com/computing/274650-the-worst-cpus-ever-made

I put my .02 on the article... I'd be happy if you guys want to contribute your .02 as well.  Bottom line is that the 9900 doesn't really belong in his aritcle.

[+OLD CS1]

4 hours ago, acadiel said:

Bottom line is that the 9900 doesn't really belong in his aritcle.

Agreed.  Talking about CPUs pushing new technologies and falling flat is fine.  But then to shoe-horn in at some random spot a CPU, which was already established at the time, as a "failure" for not being selected for a new generation of personal computers.  Not because it was a gamble on new methods or technology, but because IBM looked past it.

 

Either Joel Hruska is not a good tech journalist or he suffered a serious lack in judgement by producing this article.

 

[1980gamer]

I never owned the "worst" cpu.  But I did software support at the time.

This chip would not run many mainstream applications at all.  Not slowly, I mean, NOT AT ALL!

I supported MSSQL 7.0 etc.  It wouldn't even install, never mind trying to run it!  LOL

 

Good choice for worst!

 

I didn't realize how close TI was to dominating the CPU market!  Well, at least they were considered by IBM.

[Retrospect]

The whole "worst CPU" thing is meaningless.  CPU's are best used for the hardware they were designed for.  I've no doubt in my mind at all that the 9900 CPU will perform very satisfactory in the right hardware environment.  Didn't they use them in some hardware to run computers in Banks?  

 

It's a bit like saying the Z80A is crap, just because the Motorola 68000 gets jobs done a little quicker.  Meaningless!

[1980gamer]

You all read the article right?  The Cyrix CPU was the worst.  I agree.  It was priced well,  kind of...  It doesn't matter how little you pay for a screw driver, if it doesn't drive a screw!

The TMS9900 only failed to make it in to the IBM PC of 1980-81? when it was a 5-6 years old at the time. Really older as it was just an enhanced 960 cpu.  Not a failure, just aged out really.

 

ALL CPU's are failures compared to anything a decade newer.  CPU's until recently doubled in speed ( instructions per cycle + clock speed ) every 18 months.  Not talking just clock speed.

The further back you go, the greater the gap was.  Now, we add threads to get increased processing power.  The clock speed race came to an end thankfully. As it was not really a good metric.

 

And to be even more fair, IBM wasn't thrilled with the Intel CPU, thus you could add a math co-processor. $$$$$$$

 

But I totally agree.  Most CPU were designed for a specific device.  If someone used it for another device... That was Gravy!  

Many CPU that ended up in a "computer" or gaming console were built for military devices first.

The 6502 had great legs!  Used way beyond it's time IMO.  Granted I am counting the 6510 etc. in the family.

 

The Z80 was the first CPU I ever wrote assembly on.  It was the CPU used when I was in electronics school in the 80's

Wish I stuck with assembly instead of falling back into BASIC and C.

 

[+OLD CS1]

Cyrix the worst?  Depends upon your perspective.  When I ran the computer lab at our local college, I would come in first thing in the morning and turn on all the machines to warm up the lab.  Once it reached a comfortable temperature I would turn off machines to maintain.

 

Worst CPU, indeed.

[JB]

Cyrix was the worst if you were playing Quake. 

 

 

Meanwhile, it seems like most of the world's perception of the 9900 is shaped by the 99/4a(and all its quirks) and just a handful of sources, like https://spectrum.ieee.org/tech-history/heroic-failures/the-inside-story-of-texas-instruments-biggest-blunder-the-tms9900-microprocessor 

 

And certainly, the 9900 had a few issues. So did the 8086.

 

As I understand things, the big thing that tipped IBM to Intel's offering was a non-technical feature: Intel was JUST a chip maker. TI, by contrast, made computers, and were one of the biggest names in the industry.

This had a few repercussions for a 9900-based PC. One was that it put you on the back foot because you were buying parts from your biggest competitior. The other was that TI kinda didn't want to SELL you parts all that badly because you would be competition.

 

Intel offered all the developer support you could ask for in 1979, because they really really wanted you to buy their parts, and this support came without any baggage or ill will. TI's external developer support was somewhat miserly, to my understanding.

 

And it wasn't like TI hadn't JUST yanked the rug out from under all their calculator customers(hi, Commodore fans!) by putting the Datamath on sale for less than they had been selling the components. Trusting them to not do it again would be hard, and I imagine IBM's team was already looking for the knife.

[wierd_w]

The cyrix also had a NASTY reputation of having a rinky-dink sized on-die cache size, in addition to its poor integer math performance.

 

This translated to quite a few performance issues when running win9x, which made excessive use of context switching, and thus would suffer issues from cache exhaustion. Since systems that had cyrix cpus often had little or no external cache to speak of, the problem was greatly compounded. (See also, Packard Bell with Cyrix CPU.)

 

This is not to say that a system with a cyrix cpu was guaranteed to be a potato; Just that the circumstances leading to the use of a cyrix CPU, often resulted in a slew of additional contributory factors, ultimately resulting in peak potato-ness 9/10 times.