• Please review our updated Terms and Rules here

So how many cores are enough these days?

Agent Orange said:
I said "one" of the most successful not "the' most.
Click to expand...

I know, that's literally what I said, do you even read? "While the 386DX-40 certainly was among the more successful CPUs, I highly doubt it's one of the most successful in terms of sales."
The logic behind that is simple: "One of the most successful" would mean it's what... in the top 5? Top 3 even? I doubt it.

Agent Orange said:
It would have been real hardship and almost cost prohibitive for the up and coming enthusiast to afford a 486 in the early going.
Click to expand...

Early going was 1989, for the 486. The 386DX-40 didn't arrive until late 1991, by which time there were various different 486es, including the SX models that were certainly affordable. Remember, Pentium already arrived in early 1993.

Agent Orange said:
it just wasn't within reach at that time.
Click to expand...

Not for everyone perhaps... Then again, 386 was quite expensive as well. In 1991 I still had a 386SX-16 system. I got my 486DX2-66 in 1993 or 1994 I think (prices had dropped considerably by then, because of the new Pentium, and AMD's 386 and 486 CPUs competing with the 486SX/DX lines). The few friends that got a 386DX-40 system got theirs around the same time as I got my 486, or even later.
Because don't forget... even though the 386DX-40 was released in late 1991 on paper, I didn't actually see them in stores until much later, at which time I was already planning a 486DX2-66 (the upgrade from 386SX-16 to 386DX-40 was not that spectacular, unlike a 486DX2-66 VLB system). Supplies were apparently extremely low in the early days.
 
Scali said:
They didn't do too well financially...
Click to expand...

*snicker* Yeah, just trying to take the edge off. I unfortunately own one of their last consumer-targeted products, a C7-M in an HP 2133. (Which is a beautiful piece of hardware, far and away one of the prettiest first-gen "NetBooks" ever made, but the CPU pretty thoroughly ruins it; a well-tuned Pentium II from ten years earlier has a good chance of outrunning it.) Last I remember they were really trying to push the "Nano" platform as an Atom competitor but I've never seen it on anything but embedded/itx boards. They even make a quadcore one, which I would almost want to play with just for the heck of it if it wasn't for the whole "you *will* be disappointed" thing.
 
AMD 386-DX40, exceptional long running chip for its time. When paired to a high end mainboard that supports it, meaning proper cache, and if you can it, vlb on 386, this chip can compete with any 486 up to a dx33 or 40. It starts to fall behind big time on the 486-66 and up.

80486 and variants - has, had, and always will outperform the 386,n except the slower sx-chips, even the faster sx chips were ok, depending on what you fed them.

80486 DX-50 I have one and it is REALLY REALLY finicky beast, but when everything works properly at 50mhz, this one can eat dx-66s and maybe even 80s for breakfast.

Now please, this thread is starting to become a war of words. and quite frankly, that's not why i read these threads. Intel produces faster cooler running chips. AMD chips run far hotter, in my experience, always has been the case. Can we put this divergence of subjects to rest. That is what the off topic category is for.
 
Caluser2000 said:
For the record I didn't quite get AOs comparison between the 386DX40 or the 468DX50 either.
Click to expand...

Being that the AM386/40 version (not Intel's) was preceded by the 486DX50 there are existing articles (Google pls) on the subject. I never stated that the 386 could out perform the 486, anyone knows that would be silly. What I did say was it could hold its own for comparison sake.
 
k2x4b524[ said:
vlb on 386
Click to expand...

As far as I know, that is not possible, since VLB connects directly to the 486 bus. Pentium boards with VLB are rare as well, and generally don't work properly, because it requires some nasty kludges.

k2x4b524[ said:
except the slower sx-chips
Click to expand...

The 486SX is not a slower chip. It's the same chip, with the FPU disabled. As long as you run software that doesn't use the FPU (which most software didn't, in those days, since FPUs were an expensive add-on, which not many people invested in), there is no difference with a 486DX at the same clockspeed. A 486SX uses the exact same socket, chipset, caches and localbus as a 486DX.
 
Agent Orange said:
What I did say was it could hold its own for comparison sake.
Click to expand...

For some definitions of "hold its own" perhaps...
As I've shown with the article I linked, the 486SX-25 was generally 10% faster than a 386DX-40.
So what do you think will happen if you pit that same 386DX-40 against a 486DX-50, which not only runs at twice the clockspeed, but also has its FPU enabled?
The 386DX-40 doesn't stand a chance.
 
They do exist, i had one for a time that i was trying to find a chip for. I believe i even asked a question about it on this forum about such.

Scali said:
As far as I know, that is not possible, since VLB connects directly to the 486 bus. Pentium boards with VLB are rare as well, and generally don't work properly, because it requires some nasty kludges.

While that may be true, i speak from experience on this one. I guess it really comes down to the hardware that they are installed with.

The 486SX is not a slower chip. It's the same chip, with the FPU disabled. As long as you run software that doesn't use the FPU (which most software didn't, in those days, since FPUs were an expensive add-on, which not many people invested in), there is no difference with a 486DX at the same clockspeed. A 486SX uses the exact same socket, chipset, caches and localbus as a 486DX.
Click to expand...
 
Scali said:
For some definitions of "hold its own" perhaps...
As I've shown with the article I linked, the 486SX-25 was generally 10% faster than a 386DX-40.
So what do you think will happen if you pit that same 386DX-40 against a 486DX-50, which not only runs at twice the clockspeed, but also has its FPU enabled?
The 386DX-40 doesn't stand a chance.
Click to expand...

It's apples and oranges. You are using benchmark lab figures to justify all of your arguments. What people are referring to is overall performance. Are you going to tell us that a 486/50 is going to run WordPerfect 4 or 5 twice as fast as a 386/40 based on clock speed? Set them both side by side and you'll hardly notice the difference in real time.

We've stolen this thread, and I for apologize to the OP for my part in it. So, I'm opting out of this one. Feel free to initiate a new thread if you wish.
 
Of course there were 486DX/sx systems/mobos without vlb something somebody seems to be conveniently forgetting. VLB slots vary between non, one, two or three on the 486 systems I have here.

As mentioned previously there are 386/486 capable vlb mobos. The OPTI-495SLC, whose manual I have sitting on my lap, is one such motherboard. The owner originally bought the board, I have the original purchase documents(1993), with a DCL33 processor (have that as well) then later replaced that by upgrading it to a 486DX2/66 cpu a couple of years later. Interesting thing is the manual specifically states which slot should be used for the video card and which should be used for the ide/i/o controller. Funnily enough the system with this mobo is in was put together by the same now defunct clone builder I got my original 286/16 system from in 1990.

There's nothing particularlly odd about threads going OT. It happens here all the time.
 
Last edited:
Yea, this thread has gone all over the place, but its 14 pages of interesting so far.

I have a 386/486 VLB board but the VLB slot was never soldered in. I figured if you used the 486 the VLB slot work, with a 386 you were stuck with ISA.

Anyway my motherboard with the 2 x 4 core XEONs will be here tomorrow I think, FBDIMMs arrived today, so I have something to play with.

What is the deal with FBDIMMs anyway, pretty much just XEON processors need it?
 
Agent Orange said:
It's apples and oranges. You are using benchmark lab figures to justify all of your arguments.
Click to expand...

Not exactly. I'm using ratings of real-world programs.

Agent Orange said:
What people are referring to is overall performance.
Click to expand...

We're talking late 80s/early 90s... The performance of a single application *is* overall performance. We didn't have multitasking OSes yet, we didn't have virusscanners in the background, we didn't have virtual memory or anything. Life was simple back then.

Agent Orange said:
Are you going to tell us that a 486/50 is going to run WordPerfect 4 or 5 twice as fast as a 386/40 based on clock speed? Set them both side by side and you'll hardly notice the difference in real time.
Click to expand...

Oh please, has it come to this?
We are discussing the speed of the hardware. Choosing software where the speed of the hardware is irrelevant to the user experience is a huge cop-out. The sheer scale of this broken logic just baffles me.
As I already said, I'm an engineer, I just want to measure what the CPU can do. I'm not interested in how this may or may not suit random users' requirements. This is not buyer's advice, this is being interested in the state-of-the-art of technology. Do you understand the difference?
 
Last edited:
Caluser2000 said:
Of course there were 486DX/sx systems/mobos without vlb something somebody seems to be conveniently forgetting.
Click to expand...

I am not forgetting that at all, and certainly not because of 'convenience'. First generation 486 was ISA-only, everyone knows that.
Why do you keep trying to paint me as pro-Intel?
As a matter of fact, one of my first 486DX2-66 systems actually was an Am486. It's not about the brand, it's about the fact that the 486 was a newer generation of hardware, with considerably better performance than the 386 platform (again, not about brand, Intel and AMD 386 platforms were identical, as were their 486es in the early days).

If I was a 'fanboy' of anything, it is a fanboy of technology and performance. Which is why I liked the Amiga so much. Not because it happened to be Commodore or whatever, but because it was a very well-designed platform, doing things that none of its contemporaries were capable of. The brand is irrelevant to me (in fact, in the case of the Amiga, one could argue that it's not really a Commodore at all. It was developed by Hi-Toro, a company founded by former Atari employees. It was more Atari than Commodore in a way, oh the sweet irony). Getting attached to a particular company is irrational. Liking a certain product of a random company because the product has unique and impressive characteristics, that's rational, see the difference?

You also have to put things in the proper perspective... In the 386-era, the late 1980s/early 1990s, us Amiga guys were still laughing hard at how inept PCs were at anything. The 486 was the first PC we had to take seriously, because it had enough brute force to do pretty much everything that the Amiga custom hardware could do, and it could actually do things that Amigas could not. 386 and 486 is a world of difference.
 
Last edited:
Scali said:
386 and 486 is a world of difference.
Click to expand...
My own C code compiled on SCO UNIX in the early 90's ran almost exactly twice as fast on a 486DX/33 as on a 386DX/33. Both were a lot faster than my PDP-11/73... Or a MicroVAX 3100/20.
 
From now on the only acceptable replies to this thread are unsigned integers directly responding to the question "So how many cores are enough these days?".

2.
 
My main computer has a Core i7-4930k with 6 hyperthreaded cores (12 logical) and like others are saying, the main things that will tax it to 100% are video encoding, software development, and they are also handy for VMs. It blows away my old machine, an AMD FX-8150 8-core. It's not even close in performance, even with 2 fewer cores! Most things are twice as fast, at worst. I'm never going AMD again unless something really changes in their designs.
 
images
Eudimorphodon said:
From now on the only acceptable replies to this thread are unsigned integers directly responding to the question "So how many cores are enough these days?".

2.
Click to expand...
I think you're off the mark by 40.
 
You must log in or register to reply here.
Back
Top