Can Win-98 benefit from CPU with hyperthreading?

[herbalist]

According to Herbalist, the answer is zero.

Where did I say zero?

[dencorso]

OK.  Then, when herbalist said:

 

Most of the applications that could benefit from it won't run on 98 anyway.

 

He left unsaid: "then again, those few which could benefit from it and actually manage to run on 9x/ME, will also reap no benefit whatsoever, because, in any case, the OS is single-threaded, and therefore all threads created must execute one at a time, and never truly in parallel.

 

Well I can't exactly agree (or disagree) with what he left unsaid. So why put words into his mouth?

I did not put any words in herbalist's mounth! When *I* told you what he left unsaid, *I* told it to you from *my own* mounth!

 

And what I told you is: no mater whether one writes any software to benefit or not from multi-threading, and regardless of whether such software was also written to be compatible with 9x/ME or not, if ever anyone coaxes the said software to work on 9x/ME, it won't reap any benefit whatsoever from a hyper-threading processor *because* 9x/ME are single-threaded. 

 

[rloew]

Writing Multi-Threaded Software allows the program to take advantage of Multiple Cores when run under a Multi-Core aware OS. The same Software can run under Windows 9x but the speed advantage will be lost.

If such a program cannot run under Windows 9x, it was explicitly designed to require Multi-Core or uses other features that Windows 9x does not have. The reason there are so few programs is because the writers chose to abandon Windows 9x.

[jaclaz]

Conversations:

 

Program: Hey, OS are you there?

OS: Sure I am here, ready.

Program: Which OS are you?

OS: You should be smart enough to discover it by yourself, shouldn't you?

Program: Hmmm, OK, but are you multi-core aware?

OS: Probe me and you will see.

Program: I see that you are Windows 98 and not multicore-aware, which is a pity as I have been written with multi-core support.

OS: Well, but surely you have single core capability, don't you?

Program: Sure, now that you tell me this I could have it, but I was compiled without support for it, I need a multicore-aware OS.

OS: Sorry but you cannot run, then.

 

jaclaz

[Nomen]

Program: Hey, OS are you there?

OS: Sure I am here, ready.

Program: Which OS are you?

OS: I'm Windows XP.

Program: Hmmm, OK. How many cores are available to you?

OS: Probe me and you will see.

Program: I see that you are running on a system that has single-core CPU which is a pity as I have been written with multi-core support.

OS: Well, but surely you have single core capability, don't you?

Program: Sure, now that you tell me this I could have it, but I was compiled without support for it, I need a multicore-CPU.

OS: Sorry but you cannot run, then.

So tell me how common the above scenario is. It certainly would have been VERY common for XP to be running on a single-core CPU during it's first 4 or 5 years of life. Many budget computers were made with single core Pentiums and Celerons.

I'm not questioning or debating the fact that software that defacto MUST have a multi-core CPU available to it can therefore not run under 9x/me.

I'm questioning the assertion (that this thread has turned into) that not having multi-core CPU support is some sort of fundamental show-stopper for win-98. I think we all know that quite a lot of software that is multi-core "aware" is also single-core compatible, and hence the operatibility of that software on win-98 will depend on other factors other than 98's lack of multi-core support.

[jaclaz]

I'm questioning the assertion (that this thread has turned into) that not having multi-core CPU support is some sort of fundamental show-stopper for win-98.

The only issue with this being that no such particular assertion has been made on this thread.

 

I think we all know that quite a lot of software that is multi-core "aware" is also single-core compatible, and hence the operatibility of that software on win-98 will depend on other factors other than 98's lack of multi-core support.

Sure , the multi-core is the "last and least" of the issues with software not designed with Windows 9x compatibility in mind.

 

So tell me how common the above scenario is.

 

List:

• software that was written to be compatible with single core ONLY AND compatible Windows 9x OS
• software that was written to be compatible with single core AND multi-core BUT NOT with Windows 9x OS
• software that was written to be compatible with single core AND multi-core AND THAT (by pure chance) is also compatible  with Windows 9x OS
• software that was written to be compatible with single core BUT NOT with Windows 9x OS
• software that was written to be compatible with single core AND multi-core AND Windows 9x OS
• software that was written to be compatible multi-core  ONLY AND ( implicitly) NOT Windows 9x OS)

#1-2 Common

#3-4 Fairly common

#5-6 Less common

 

jaclaz

[herbalist]

I don't install or try out much new software, but haven't run into any software that's failed to run on 98 due to a lack of multi-core support. For me, it's almost always compatibility issues that KernelEx can't fix, such as:

Any version of SeaMonkey past 2.0.14

Any version of Tor past 0.2.3.25

Current modifications to Proxomitron that address filtering HTTPS

Any applications that rely on the current versions of OpenSSL such as ProxHTTPSProxy.

For me, there's no single show stopper. There's a growing number of incompatibilities that I can't compensate for. I didn't regard using an older browser as a security/privacy risk as long as Proxomitron was filtering the content. With Proxomitron unable to filter the current HTTPS, the browser is exposed and is becoming increasingly unsafe to use. With Tor, it won't be long before version 0.2.3.25 won't be allowed to serve as a relay or exit and may not be accepted as a client. When that happens, I either have to:

1, Install linux and learn to secure it well enough to serve as an exit,

2, Shut the exit node down.

Slowly the decision is being made for me.

[dencorso]

For me, there's no single show stopper. There's a growing number of incompatibilities that I can't compensate for.

 

That's how I see it right now for 9x/ME and somewhere in the near future (but not quite right now) for XP SP3.

[herbalist]

If I had to guess, XP will be maintained (unofficially) longer and better than 98 was. It has a lot bigger user base. Although it's different than Win 7, 8, 10, etc, it's not as different as the 9X systems. 98 supplied the proof that an OS can be maintained past Microsofts so-called "end of life" and be made better than anything they ever did for it. 98 was used to establish the methods and create the tools, tools I expect should see a lot more development on XP. 9X and its users proved that "end of support" is not end of life.

I will migrate to XP if I'm forced to, but I'll need better hardware to get the same amount of work out of it. I tried using XP to run the Tor relay on this PC. It rarely lasted a full day. With the same hardware running Lite98SE with the modifications I've found here, it would last 2 weeks or better before needing a reboot. AFAIC, XP can be made reasonably secure with a lot of stripping and service disabling, but it's always felt like I had to fight the OS to get what I wanted from it, or more accurately, to get rid of everything that I don't want. With 98, it did what I wanted and only what I wanted, then stayed out of the way. I wish MS would get off of this twisted "computing experience" concept of theirs. Ones computing experience shouldn't be influenced by the OS. People don't use operating systems. They use programs that run on that operating system. Microsoft don't seem to know the difference between an interface and an obstacle.

[dencorso]

If I had to guess, XP will be maintained (unofficially) longer and better than 98 was. It has a lot bigger user base.

 

Sure. Then again, I don't think MS was ever so keen on getting rid of 9x/ME (especially 98SE) as it is right now on getting rid of XP...

[herbalist]

Then again, I don't think MS was ever so keen on getting rid of 9x/ME (especially 98SE) as it is right now on getting rid of XP...

 

They got pretty extreme about it with 98 too, just in a different way. With 98, it seemed their biggest weapon was deliberately introduced and artificial incompatibility. With XP, it's extreme rhetoric regarding insecurity, hacker heaven, etc. Maybe that because it'll be easier to make newer software work on XP than it is to make it run on 98.

IMO, XP is the last version of Windows that a user can really control. On XP, all of the logging/babysitting features can be disabled. With some effort, all of the open ports can be closed. As far as I'm aware, it's impossible to close all of the ports on Vista and newer systems. It's also impossible to prevent them from calling home without using a separate device between them and the internet. AFAIC, the current operating systems are little more than bloated spyware coated in eye candy.

[rloew]

To be even more precise, THE Hyperthread in THE Core of THE CPU handed to it by the BIOS.

Normally this would be the BP (Base Processor) designated by the Hardware.

Updating what I said previously, Windows 9x runs in the Core handed to it by DOS, not necessarily the BIOS.

I am writing this on a Windows 98SE System running from the second Core, an AP (Application Processor) not the BP.

[CamTron]

Correct me if I'm wrong, but I always thought that the way to make your program multi-core capable was to use multiple threads, and if the OS is smart, it would divide the threads of the application between cores so that they efficiently run in parallel. On a single-core CPU, all threads would just run on the same core, and the OS would pre-empt them like it does with separate application. Thus, unless you're deliberately trying to prevent your application from running on a single-core CPU, it should still run, albeit not as efficiently.

Edited December 9, 2014 by CamTron