The Solutor wrote on Yesterday, 00:18:

Any athlon 64 / opteron released in 2003) supported more than 3.3 GB (that is the practical limit of XP 32 NOT 4GB), given the memory controller is in the CPU, not the MB

You just walked right into your own trap.

1. **Consumer Athlon 64 in 2003 was Socket 754:**
When the Athlon 64 launched in September 2003, desktop consumers bought Socket 754. Socket 754 had a single channel memory controller and motherboards came with 2 or 3 DIMM slots. Since consumer unbuffered DDR1 topped out at 1 GB per stick, a Socket 754 board had a physical ceiling of **2 GB or 3 GB**. You physically could not install >3.3 GB into a consumer desktop Athlon 64 in 2003.

2. **Opteron in 2003 was Socket 940 enterprise server hardware:**
The only K8 platform in 2003 that could physically exceed 4 GB was Socket 940, which required **Registered ECC server RAM** on multi thousand dollar server/workstation boards (Tyan Thunder, Iwill). You spent the last three days mocking enterprise hardware as "mambojambo that small businesses never used" and now, to name a 2003 CPU that could exceed 3.3 GB, you have to cite an enterprise server Opteron running server Registered ECC RAM.

3. **The motherboard BIOS still controls memory remapping:**
Even with the memory controller on the CPU die, the K8 IMC doesn't live in a vacuum. It routes through the HyperTransport bus and relies on the motherboard BIOS to construct the ACPI e820 address map and set the base/limit registers for DRAM hole remapping. On 2003/2004 nForce3 150/250 and VIA K8T800 boards, BIOS remapping was notoriously broken or non existent, leaving MMIO space trapped below 4 GB regardless of the CPU.

Once again, you are conflating enterprise server hardware (Socket 940 Opteron) with the consumer desktop reality of 2003–2004.

The Solutor wrote on Yesterday, 00:21:

ElectroSoldier wrote on Yesterday, 00:20:

Once you get to that hard stop that there were no consumer motherboards that supported more than 4Gb RAM you arrive at the next theory killer.

You are late.

You too. I was expecting the K8 reply just after the list of chipsets that didn't support >4gb RAM.

ElectroSoldier wrote on Yesterday, 02:19:

2. **Opteron in 2003 was Socket 940 enterprise server hardware:**
The only K8 platform in 2003 that could physically exceed 4 GB was Socket 940, which required **Registered ECC server RAM** on multi thousand dollar server/workstation boards (Tyan Thunder, Iwill).

1) you don't need to "exceed" 4GB, you need to REACH 4GB to be hampered by SP2

2)

support for a single channel memory controller (64 bits wide) with a maximum of three unbuffered DIMMs, or four registered DIMMs

3) DDR400 ECC 2GB modules exist https://www.amazon.com/Micron-Module-PC3200-R … ps%2C262&sr=8-1 And practically any AMD CPU support them even if not advertised

4) All of the above is irrelevant anyway, as I explained many times. The only purpose of debunking your false claims is fun.

You spent the last three days mocking enterprise hardware as "mambojambo that small businesses never used" and now, to name a 2003 CPU that could exceed 3.3 GB, you have to cite an enterprise server Opteron running server Registered ECC RAM.

Most, if not all, single socket Opterons can use non ECC RAM just like Athlons can use ECC, unless artificially (AGAIN) limited by the BIOS

And yes, that's makes the mambojambo argument even stronger.

3. **The motherboard BIOS still controls memory remapping:**
Even with the memory controller on the CPU die, the K8 IMC doesn't live in a vacuum. It routes through the HyperTransport bus and relies on the motherboard BIOS to construct the ACPI e820 address map and set the base/limit registers for DRAM hole remapping. On 2003/2004 nForce3 150/250 and VIA K8T800 boards, BIOS remapping was notoriously broken or non existent, leaving MMIO space trapped below 4 GB regardless of the CPU

Even if true, which I'm anything but sure, 4GB is still more than 3.3GB (and 3.3GB is the best case scenario, could be even 2.9GB depending the HW), still 700+MB wasted in any 32bit Windows released between 2004 and the present day, if not patched.

But feel free to think that the MS scam is just in the mind of those "hideous cospirationists", but just tell that to people ignorant like you, not to me, I can grind your flaky arguments, with incredible ease, as anyone can read.

Seriously learn something or find another hobby.

The Mandela effect has arrived in the tech space.^

Those DIMMs exist now, but at the time of XPSP2 they didnt not on consumer grade hardware.
Server hardware they did, but you have already dismissed that as not required. But then server grade hardware isnt in dispute here. (I had and still have a Supermicro P3TDE6 that supported 4Gb ECC RDIMMs PC133 but that has a ServerWorks chipset that was from 2001, but that doesnt count, its Enterprise grade and small to medium size businesses didnt use it)

It is interesting though you have conceded the fact that no 32bit hardware can do what you said it could.
And you seem now also have conceded what I said all along was true. Because your only fallback position now you have realised what the hardware of the time actually supported not what you seem to remember it supported and that enterprise grade hardware was required.

Which is what I said in the first place.

The HAL clamp is at 0xFFFFFFFF (4.0 GB).

The reason systems showed ~3.3 GB was that the motherboard MMIO ranges sat between ~3.3 GB and 4.0 GB. Because the HAL prevented remapping those physical addresses above 4 GB, the overlapping RAM was hidden.

You dont even seem to know the technicals of what you are talking about.
To anybody who knows the technical details you look like a somebody who doesnt have the first clue of what theyre talking about.

You said that yourself when you said "Even if true, which I'm anything but sure"

ElectroSoldier wrote on Yesterday, 05:56:

It is interesting though you have conceded the fact that no 32bit hardware can do what you said it could.

Yeah, keep adding hallucinations to the other you already brought

Now that I demonstrated that 4GB on a period consumer machine, you added "32bit HW "to something I never said. Just to be sure.

What a clown

The reason systems showed ~3.3 GB was that the motherboard MMIO ranges sat between ~3.3 GB and 4.0 GB. Because the HAL prevented remapping those physical addresses above 4 GB, the overlapping RAM was hidden.

We aren't discussing the reasons, that everybody knows, we are discussing the outcome for end user.

And the end user doesn't give a crap about the reasons, that RAM is unavailable. In our case is unavailable because MS stole it, which is what we are talking about, something that seem everything but clear to you.

The Solutor wrote on Yesterday, 09:04:

Yeah, keep adding hallucinations to the other you already brought […]
Show full quote

ElectroSoldier wrote on Yesterday, 05:56:

It is interesting though you have conceded the fact that no 32bit hardware can do what you said it could.

Yeah, keep adding hallucinations to the other you already brought

Now that I demonstrated that 4GB on a period consumer machine, you added "32bit HW "to something I never said. Just to be sure.

What a clown

The reason systems showed ~3.3 GB was that the motherboard MMIO ranges sat between ~3.3 GB and 4.0 GB. Because the HAL prevented remapping those physical addresses above 4 GB, the overlapping RAM was hidden.

We aren't discussing the reasons, that everybody knows, we are discussing the outcome for end user.

And nd end user don't give a crap about the reasons, that RAM is unavailable. In our case is unavailable because MS stole it, which is what we are talking about, something that seem everything but clear to you.

No you havent.
An Opteron isnt a consumer grade PC and never has been. In fact Opteron was one of those enterprise grade options you spercifically said was part of the conspiracy.
But that is the one you settled on which proves you wrong.

The Solutor wrote on Yesterday, 13:03:

Seriously your need of living in a constant dick size contest, puts everyone in doubt about your masculinity.

Using prosthetics, like you do, makes the matter even more ridiculous.

No what it comes down to now is you dont want to say it was a platform limitation not an OS limitation at the time. Which is exactly what I said several pages ago. It became a OS limitation later sure but that was later, and they had a problem with MSBlast NOW!.

I can understand a conspiracy theory about Microsoft, a company doesnt get that big without some dodgy practices in their past. I mean the way Gates swiped DOS is a good example of that, but to call into question XP SP2 considering what was happening in the computer world at the time, and what Microsoft had to do to make SP2 happen is just crazy. (it makes it look like you dont know what happened then, in which case where you even there? Because I can prove I was...)
Yes it lead to a regrettable situation with the >4Gb clamp on XP, but considering what it fixed, what people were doing to make the situation worse, what it cost the company is terms of software development capability, in that MS had to divert developers from Project Longhorn, which was dying on its feet at the time by the way and maybe be one of the reasons we only ended up with Vista the way it was, and pile those developers into the development of SP2, then on top of all that the clamp on XP meant the mass market never needed the copious amounts of RAM thus driving down the cost of RAM which Microsoft needed, and the poor state of drivers in the consumer market, but what else could they have done given the Blaster and Sasser worms among others?
The motherboards at the time simply didnt support that much RAM in the first place, the limitation was a platform one, nobody had >4Gb of RAM anyway so the clamp seems logical and Microsoft never expected XP to last anywhere near as long as it did and when it did it just became a larger and larger source of embarrassment. And one they were forced into.

Considering all of the problems they had, features needlessly cut it is strange you settle on this one as a hill to die on in a forum like this. All it looks like for you now is an optical juggle for you to not walk away from it looking like you have no idea of the hardware from the time, which makes me question your real world experiences, especially as it all seems to come from running VMs today.
How many off ramps do you want me to give you ?

They took a shortcut, they could have enforced PAE test on the WHQL certification, but they preferred to throw the baby out with the bathwater, pushing the x64 thing that was (and mostly still is) detrimental rather than an improvement for most users.

That is, not the dirtiest MS move ever, but still dirty enough.

The Solutor wrote on Yesterday, 13:44:

They took a shortcut, they could have enforced PAE test on the WHQL certification, but they preferred to throw the baby out with the bathwater, pushing the x64 thing that was (and mostly still is) detrimental rather than an improvement for most users.

That is, not the dirtiest MS move ever, but still dirty enough.

Willing sacrafice a feature they thought wasnt going to be important in the future of that OS I will agree with. A shortcut... I wont go that far.

Longhorn was already in Beta stage, I still have the files with their creation dates 4051 25/03/04, 4053 21/03/04, 4074 04/05/04. I didnt get them 0day ts true but I remember using them before SP2 went final.

Oh, sure! How can be important to support an OS that survived for another decade after the SP2 release in the server flavour, and another six years in the client variant?
How can be important to support XP when Vista was released in a form that defining Alpha would be a compliment?

Aside me that got a living upgrading machines from Vista to XP and from 2008 to 2003, Lenovo, HP , Toshiba made a ton of money overpricing their "pro" lines for years, where "Pro" was intended as the ability to upgrade from Vista to Xp, officially, w/o violating the license.

X64 is not perfect, no, but still better for users than Intel's Itanium was.

PAE does not really have a strong reason to exist with flat native access.

Much like there's no real need for EMS paging.

Would it be nice to have fixed by the community? Absolutely. Will MS ever do that? Pshaw! No!

Arguing over water that passed under the bridge literally over a decade ago is silly, IMO.

MS really should have had stronger words with driver makers, that 32bit was on the last miles of its journey, and they needed to focus on >32bit address space awareness and on behaviors needed to conform with that future.

They did not. That's why so many HDA solutions just wont turn on with >2gb installed. Even on XP64, and even on win7. I have umbrage, much like many others do. My being angry wont fix it.

For devices that could support it, PAE would let you get multitasking benefits of more ram being present. I've seen the argument made that it does not let proceses use more than a 32bit address space allows within a single process, like that's the only thing that matters, and that thirsty background processes (like notorious antivurus and security suites) weren't things. PAE allowed more memory being present to help mitigate that resource hit, when they'd inevitably want to start scanning or updating for no good reason.

Instead of your game suddenly having multiple hundred megabytes allocated away from its visible pool, and it having to adapt by lowering lod and cutting render quality, PAE gave that shitty garbage AV its *OWN* 32bit address space to hog-wallow in. Much better.

A very welcome feature, for 32bit machines.

wierd_w wrote on Yesterday, 14:06:

For devices that could support it, PAE would let you get multitasking benefits of more ram being present. I've seen the argument made that it does not let proceses use more than a 32bit address space allows within a single process, like that's the only thing that matters, and that thirsty background processes (like notorious antivurus and security suites) weren't things. PAE allowed more memory being present to help mitigate that resource hit, when they'd inevitably want to start scanning or updating for no good reason.

A very welcome feature, for 32bit machines.

Window update alone, in Vista x86 takes so much memory that is almost impossible to run it w/o PAE patch and 4+ GB of ram

The Solutor wrote on Yesterday, 14:06:

Oh, sure! How can be important to support an OS that survived for another decade after the SP2 release in the server flavour, and another six years in the client variant?
How can be important to support XP when Vista was released in a form that defining Alpha would be a compliment?

Aside me that got a living upgrading machines from Vista to XP and from 2008 to 2003, Lenovo, HP , Toshiba made a ton of money overpricing their "pro" lines for years, where "Pro" was intended as the ability to upgrade from Vista to Xp, officially, w/o violating the license.

It wasnt Microsofts intention for it to last another 6 years. The server OS is another matter, which had its own set of problems thats true.
By the time SP2 went final XP was already 3 years old or 2years 10 months, something like that.
It was extremely old considering the march of technological advancement. Win98SE barely made it to a year before it was replaced by WinME... But we wont talk about that! 😀

Do you remember how Microsoft was pressured into continuing support for XP in spite of clearly wanting to kill it dead?
external developers wouldnt leave it alone, they wouldnt port apps to x64 (most of them clearly didnt know how to when you look at the software they did turn out) and they certainly couldnt deal with multi threading apps despite SMP being a thing in the workstation market for years.
I know XP is much loved by us. Im like you, I made a lot of money out of the XP over Vista wars. But to M$ it was a source of problems and in the end embarrassment.
Technology left XP behind not long after SP2 went final. As has been noted here it wasnt long, time wise at least, until you got the first motherboards that did support >4Gb of RAM and not long after people did feel the 4Gb clamp on the HAL. Thats a real problem for a company whos public see their flagship flounder (Vista) and pefer the old faithful as the go to ship to sail on.
It wasnt rained in until Win7 for M$, and even then XP was still there, a corps of its former self but still there being replaced by 7. Thats a definate because there were so many offices here that went from XP to 7, years after those XP machines should have been ewaste.
XP provided to downgrade path to XP because of demand, not because they wanted to.

By the time XP was in its twilight years driver developers had learnt to code for it and they didnt want to let go of it to develop software in a new arena, especially as when you look at the software you have no choice but to realise they were no good at it.

The Solutor wrote on Yesterday, 14:10:

wierd_w wrote on Yesterday, 14:06:

For devices that could support it, PAE would let you get multitasking benefits of more ram being present. I've seen the argument made that it does not let proceses use more than a 32bit address space allows within a single process, like that's the only thing that matters, and that thirsty background processes (like notorious antivurus and security suites) weren't things. PAE allowed more memory being present to help mitigate that resource hit, when they'd inevitably want to start scanning or updating for no good reason.

A very welcome feature, for 32bit machines.

Window update alone, in Vista x86 takes so much memory that is almost impossible to run it w/o PAE patch and 4+ GB of ram

Yeah see my experience of Vista was on a dual Clovertown system with 16Gb of RAM... And it didnt last long before I went to Win7, but I know theres no smoke without fire and it was a right old dumpster fire of an OS and when you look at the plans for Longhorn and its roadmap at the time you can clearly see Vista what was meant to be and what was are very different things.
But I have no way to defend Vista at all. Not from Microsoft and not from the developer side either.

wierd_w wrote on Yesterday, 14:06:

MS really should have had stronger words with driver makers, that 32bit was on the last miles of its journey, and they needed to focus on >32bit address space awareness and on behaviors needed to conform with that future.

They did not. That's why so many HDA solutions just wont turn on with >2gb installed. Even on XP64, and even on win7. I have umbrage, much like many others do. My being angry wont fix it.

Spot on, especially regarding High Definition Audio.

That is the exact point I’ve been making throughout this entire thread: the real bottleneck was never Microsoft artificially holding back the client OS out of spite, but peripheral vendors cutting corners on their DMA controllers and driver address descriptors. Creative, Realtek, C-Media, and budget PCI card makers built hardware for an ecosystem where 256–512 MB was standard, and when faced with high physical addresses, their drivers simply fell apart.

Microsoft's mistake wasn't clamping XP SP2 to 4 GB in 2004—that was necessary triage to keep machines from blue screening once DEP forced PAE mode active. Their mistake was not cracking down on third party WHQL driver validation years earlier. They finally enforced mandatory 64bit driver signing with Vista x64, but by then, the 32bit consumer driver landscape was already the Wild West.

And you're entirely right about PAE multitasking. People get fixated on the per-process 2 GB/3 GB virtual limit, forgetting that having >4 GB of physical RAM allowed background bloat (like bloated AV suites of the era) to run in their own address spaces without page swapping active games to disk.

A level-headed post and a breath of fresh air.

ElectroSoldier wrote on Yesterday, 14:45:

It wasnt Microsofts intention for it to last another 6 years. The server OS is another matter, which had its own set of problems […]
Show full quote

The Solutor wrote on Yesterday, 14:06:

Oh, sure! How can be important to support an OS that survived for another decade after the SP2 release in the server flavour, and another six years in the client variant?
How can be important to support XP when Vista was released in a form that defining Alpha would be a compliment?

Aside me that got a living upgrading machines from Vista to XP and from 2008 to 2003, Lenovo, HP , Toshiba made a ton of money overpricing their "pro" lines for years, where "Pro" was intended as the ability to upgrade from Vista to Xp, officially, w/o violating the license.

It wasnt Microsofts intention for it to last another 6 years. The server OS is another matter, which had its own set of problems thats true.
By the time SP2 went final XP was already 3 years old or 2years 10 months, something like that.
It was extremely old considering the march of technological advancement. Win98SE barely made it to a year before it was replaced by WinME... But we wont talk about that! 😀

Do you remember how Microsoft was pressured into continuing support for XP in spite of clearly wanting to kill it dead?
external developers wouldnt leave it alone, they wouldnt port apps to x64 (most of them clearly didnt know how to when you look at the software they did turn out) and they certainly couldnt deal with multi threading apps despite SMP being a thing in the workstation market for years.
I know XP is much loved by us. Im like you, I made a lot of money out of the XP over Vista wars. But to M$ it was a source of problems and in the end embarrassment.
Technology left XP behind not long after SP2 went final. As has been noted here it wasnt long, time wise at least, until you got the first motherboards that did support >4Gb of RAM and not long after people did feel the 4Gb clamp on the HAL. Thats a real problem for a company whos public see their flagship flounder (Vista) and pefer the old faithful as the go to ship to sail on.
It wasnt rained in until Win7 for M$, and even then XP was still there, a corps of its former self but still there being replaced by 7. Thats a definate because there were so many offices here that went from XP to 7, years after those XP machines should have been ewaste.
XP provided to downgrade path to XP because of demand, not because they wanted to.

By the time XP was in its twilight years driver developers had learnt to code for it and they didnt want to let go of it to develop software in a new arena, especially as when you look at the software you have no choice but to realise they were no good at it.

The Solutor wrote on Yesterday, 14:10:

wierd_w wrote on Yesterday, 14:06:

For devices that could support it, PAE would let you get multitasking benefits of more ram being present. I've seen the argument made that it does not let proceses use more than a 32bit address space allows within a single process, like that's the only thing that matters, and that thirsty background processes (like notorious antivurus and security suites) weren't things. PAE allowed more memory being present to help mitigate that resource hit, when they'd inevitably want to start scanning or updating for no good reason.

A very welcome feature, for 32bit machines.

Window update alone, in Vista x86 takes so much memory that is almost impossible to run it w/o PAE patch and 4+ GB of ram

Yeah see my experience of Vista was on a dual Clovertown system with 16Gb of RAM... And it didnt last long before I went to Win7, but I know theres no smoke without fire and it was a right old dumpster fire of an OS and when you look at the plans for Longhorn and its roadmap at the time you can clearly see Vista what was meant to be and what was are very different things.
But I have no way to defend Vista at all. Not from Microsoft and not from the developer side either.

Windows 7 is Vista, Vista Ultimate with updates on Win7 hardware performs and behaves exactly as well as Win7 does. Vista had the unfortunate position of existing between the transition from X86 to X64 and all the bullshit that went with that transition along with the transition to a new driver model which the "cheap" hardware companies hated and many users decided was Vistas fault for the driver problems.

If Vista had come along a year later it wouldn't have gotten so much hate along with MS giving devs a year heads up to start their driver transitions and I feel Vista would have been received almost as well as Win7 was.

Now if this was ME we were discussing . .then yes there is zero defense for that OS.

Mine was less enjoyable.

While my own systems had gone from Athlon64 to intel i7, with all the benefits, *my family members* skipped the vista era like a scourge plague, *and kept old hardware* that they asked (insisted) I help them with, with win7.

Thus, my many painful memories of 'LULZ! That's legacy! Upgrade! Tee hee!' from oems, vis a vis, 'drivers on 64bit OS not respecting 64bit address space. Beligerently clinging to 32bit conventions.'

And thus my deeply set umbrage.

Reminder to keep things civil people. Disagreements and debates are fine, but unnecessary antagonism is not.

wierd_w wrote on Yesterday, 15:08:

Mine was less enjoyable. […]
Show full quote

Mine was less enjoyable.

While my own systems had gone from Athlon64 to intel i7, with all the benefits, *my family members* skipped the vista era like a scourge plague, *and kept old hardware* that they asked (insisted) I help them with, with win7.

Thus, my many painful memories of 'LULZ! That's legacy! Upgrade! Tee hee!' from oems, vis a vis, 'drivers on 64bit OS not respecting 64bit address space. Beligerently clinging to 32bit conventions.'

And thus my deeply set umbrage.

Yeah . .people blame Vista for the driver problems .. no idea why since drivers were not Microsoft's responsibility at the time. Sure this changed when MS finally decided that enough was enough and enforced compliance with the driver model but none of that was an OS caused problem.

Devs simply didn't want to comply with newer driver standards ..

Trashbytes wrote on Yesterday, 15:04:

Windows 7 is Vista, Vista Ultimate with updates on Win7 hardware performs and behaves exactly as well as Win7 does.

Obviously it isn't. The native vhds alone (which most Windows people have no idea of what they are) were just a revolution and came with Win7.

Each Windows release has brought invisible but substantial improvements, say Hyper-V and vhdx in Win8, RemoteFX in Win 7, WSL in W10 1607, WSA In early win 11, that marvel which is the deduplication in Win8, and so on. Just to name the first things that came on mind.

Don't stop to the GUI alone

Whatever, Yes Vista SP2 is usable. The point is that when Vista was released there wasn't any SP1 or SP2, people needed to work and needed to work in that moment, not hoping that, one day, a SP would have solved their problems.

Now if this was ME we were discussing . .then yes there is zero defense for that OS.

Not at all, people (nerds) hated ME mainly because the DOS was hidden, something incredibly easy to workaround using dos 8 images, that are even included in the install media.

And it got a bad name for that. That is.

ME is really what is supposed to be, a better Win98[SE].
I had clients that used ME machines for specific work that lasted for more than 2 decades W/O a single problem.

Perhaps people with short memory forget how crappy was the contemporary Win2k SP0/SP1.
Sp3/4 fixed most of the problems just like in the Vista case, and Just like in the Vista case MS was quick to replace the betas with the real deals XP and Seven.

ME wasn't the beta of anything, was the most refined OS of the Win9x family, nothing else.

Trashbytes wrote on Yesterday, 15:04:

Windows 7 is Vista, Vista Ultimate with updates on Win7 hardware performs and behaves exactly as well as Win7 does. Vista had the unfortunate position of existing between the transition from X86 to X64 and all the bullshit that went with that transition along with the transition to a new driver model which the "cheap" hardware companies hated.

If Vista had come along a year later it wouldn't have gotten so much hate along with MS giving devs a year heads up to start their driver transitions and I feel Vista would have been received almost as well as Win7 was.

You hit the nail on the head. Under the hood, Windows 7 (NT 6.1) was really just Vista (NT 6.0) with a refined shell, a retuned SuperFetch/memory footprint, and three years of mature hardware under it.

Vista took all the arrows in the back so Windows 7 could look like a savior.

The biggest disaster for Vista was the massive architectural overhaul in the driver subsystem:
1. **WDDM 1.0:** Moving away from the ancient XP Display Driver Model (XPDM) was essential for stability, but graphics vendors *especially NVIDIA* shipped notoriously broken drivers at launch. Internal Microsoft crash telemetry from 2007 showed NVIDIA drivers were responsible for nearly 29% of all Vista BSODs.
2. **Rewritten Audio Architecture (WASAPI):** Moving audio into user mode stopped rogue audio drivers from crashing the kernel, but it famously broke hardware DirectSound3D acceleration (sending Creative scrambling to make ALchemy) and caused endless latency complaints until vendors figured out how to write decent drivers for it.
3. **The "Vista Capable" Hardware Lie:** OEMs pushed Microsoft to badge cheap netbooks and budget PCs running Intel 945GM graphics and a miserable 512 MB of RAM as "Vista Capable." When users installed an OS with Aero, SuperFetch, and modern indexing services on 512 MB of RAM, the system ground to a halt.

By the time Windows 7 launched in late 2009, the ecosystem had completely changed: 64bit multi core CPUs were standard, 2 GB to 4 GB of cheap DDR2/DDR3 was in every machine, and hardware vendors had finally spent three years fixing their WDDM drivers. Windows 7 didn't succeed because it was fundamentally different under the hood; it succeeded because Vista had already done the brutal, unglamorous work of dragging the entire PC hardware industry into the modern era.

wierd_w wrote on Yesterday, 15:08:

Mine was less enjoyable. […]
Show full quote

Mine was less enjoyable.

While my own systems had gone from Athlon64 to intel i7, with all the benefits, *my family members* skipped the vista era like a scourge plague, *and kept old hardware* that they asked (insisted) I help them with, with win7.

Thus, my many painful memories of 'LULZ! That's legacy! Upgrade! Tee hee!' from oems, vis a vis, 'drivers on 64bit OS not respecting 64bit address space. Beligerently clinging to 32bit conventions.'

And thus my deeply set umbrage.

See the post above. Because yeah I know they were hard times for most which is why I think my experience on Vista doesnt count. I was running a Quadro and signed guarenteed working drivers on a system that cost well over £5000 so it doesnt count.

Re: Win8/.1

If you install a thirdparty shell (classic/openshell ftw. You should use ut even on win11, in my opinion.), it is surprisingly usable.

It does have a few driver related warts; the switch to hard-mandated driver signing meant an end to (still much needed) manual .inf doctoring to slap the stupid out of nVidia's clumsy hands, or to get intel's RST driver to install. (Many of Intel's iGPU drivers, likewise, benefit from getting a manual prodding. They are *especially* shitty about arbitrary support cutoffs, even though support is completely realistic and possible still. Video drivers for Vista/7 will *totally* work on 8/.1, but the *installer* will scream and throw a tantrum, *BECAUSE MOMMY KNOWS BEST HOW DARE YOU.*) That meant going down a rabbithole of setting up and using 'local testing' self-certificates, which was onerous and painful.

That's really not MS's fault, and more driver makers', again.

MS's bullshit contribution was the schizophrenic UI nobody but they wanted, and that they've fought to keep for 3 more windows releases since.

If they lock down on 3rd party shells, there will be violence.

(Though I already switched to linux for personal use years ago. The win11 installs are for 'support facilitation' purposes.)