Pale moon messed up looks on windows 7

[Moonchild]

Just downloaded the latest version and it works, posting from it.

Thanks for confirming!

Thanks for the info.

safe moon before installing the latest one, it still had the same problem, black window.

Yeah we had already found it was a broader issue unrelated to HWA.

[billmcct]

Glad you found it. I just installed on my Win 7 and it works with HWA on.

[Pete321]

CPU: Quad Core Q9550
GPU: Geforce GT1030 (driver 461.72)
Windows 7 version: W7 Ultimate 6.1 Build 7601: SP1 - as far as i can remember, all updates installed.

If anything else is needed, please do ask

[Moonchild]

Interesting. so far the two processors mentioned are both Core 2 variants.

[Galaxy]

One of my computers has an Intel 64-bit CPU without AVX support. I have it setup to dual-boot into either Windows 7 or Windows 10.

Here are results of my testing:

I booted into Windows 7 Professional 6.1.7601 Service Pack 1 and installed Pale Moon 33.4.0-32bit using palemoon-33.4.0.win32.installer.exe. This resulted in similar display issues as reported and as shown below.

DisplayProblem1.jpg

DisplayProblem2.jpg

I then installed Pale Moon 33.4.0 64-bit-sse2 using palemoon-33.4.0.win64-sse2.installer.exe and it runs fine without issues on Windows 7 on this computer.

After that, I booted into Windows 10 Professional Version 22H2 (Build 19045.4894). I installed Pale Moon 33.4.0-32bit using palemoon-33.4.0.win32.installer.exe and it ran fine without issues.

CPU: Intel(R) Core(TM)2 Quad CPU Q9300 with support for MMX, SSE2, SSE3, SSE4.1
GPU: ATI Radeon HD 4830 v8.970.100.9001

[Moonchild]

Thanks.

So yes, it seems to be a confirmation that the issue occurs on systems with an Intel Core 2 generation CPU, irrespective of GPU make or model.
I'm assuming these are the same CPU models that would crash using a newer version of the MSVC compiler toolchain.

Although I'd love to be able to find out exactly why this occurs, I'm afraid that would be a difficult chase; unless some people with detailed knowledge of the Intel Core 2 have an idea why this might happen? Are there known architectural problems with Core 2 Duo or Core 2 Quad CPUs?

[athenian200]

Although I'd love to be able to find out exactly why this occurs, I'm afraid that would be a difficult chase; unless some people with detailed knowledge of the Intel Core 2 have an idea why this might happen? Are there known architectural problems with Core 2 Duo or Core 2 Quad CPUs?

I don't have a deep knowledge of the Core 2 Duo or Core 2 Quad architecture, but what I do know is that systems built around those architectures gave me a lot of weird problems back in the day. I had a Core 2 Duo system fail on me after less than two years, and a Core 2 Quad system have all kinds of weird issues that were so bad I wound up parting out both systems and building a Sandy Bridge and Ivy Bridge system in the same cases. I recall software not quite doing what I expected and finding myself going back to older PCs to do certain things... overall, subjectively, I will say I seemed to have nothing but issues with that family of processors. It seemed like older hardware did some things properly that these didn't, and newer hardware also did a lot of things better and was more streamlined. That was just an awkward transitional generation in a lot of ways.

Could some AVX instructions be finding their way into the 32-bit builds in violation of spec somehow? Or is it something else that's going on here? Either way, it does seem like it's the older, sketchier hardware we expected to have issues with that is running into these problems as we try to move forwards.

[UCyborg]

Interesting, back then I had a dilemma whether to go with Intel or AMD (whether to pick one of those Intel Core 2 or one of AMD Phenom II CPUs). I went with AMD, Intel was always dominant anyway. It's not very fast, but still holding up. I haven't daily driven Windows 7 in years though...

[GStathops]

Oops .. happens here also, on Win7 fully updated 32bit machine
Upon updating (internal updater) to 33.4.0 , Pale Moon doesn't work,

Yes .. Core 2 Duo 7300 here .. that renders Pale Moon 33.4.0 unusable

Update to 33.4.0.1 works

[Moonchild]

overall, subjectively, I will say I seemed to have nothing but issues with that family of processors. It seemed like older hardware did some things properly that these didn't, and newer hardware also did a lot of things better and was more streamlined. That was just an awkward transitional generation in a lot of ways.

Interesting! So it's quite possible there is, in fact, some architectural bug in those processors that is being hit by us.

The bitter pill looking forward is, unfortunately, that if we want to remain compatible we will have to keep a frozen toolchain and SDK for 32-bit builds, and I'm not sure when we will start running into build issues as a result of that. It might be years out, though, but it's definitely going to end sometime. At that juncture we'll have to decide what to do; either unfreeze the toolchain and build 32-bit on the later toolchain and SDK, or drop support for Core 2 systems. I don't think Microsoft is going to make or keep workarounds in their compilers for a target OS that is well past EoL on hardware that is nearing 20 years old, so there won't be any solution in that corner.

[athenian200]

Interesting! So it's quite possible there is, in fact, some architectural bug in those processors that is being hit by us.

The bitter pill looking forward is, unfortunately, that if we want to remain compatible we will have to keep a frozen toolchain and SDK for 32-bit builds, and I'm not sure when we will start running into build issues as a result of that. It might be years out, though, but it's definitely going to end sometime. At that juncture we'll have to decide what to do; either unfreeze the toolchain and build 32-bit on the later toolchain and SDK, or drop support for Core 2 systems. I don't think Microsoft is going to make or keep workarounds in their compilers for a target OS that is well past EoL on hardware that is nearing 20 years old, so there won't be any solution in that corner.

Well, with the move to AVX, we already don't support Core 2 systems on 64-bit builds anyway, and honestly these compiler issues broadly seem to suggest our intuition about what hardware we can continue to support moving forward was correct. If it hadn't been the AVX issue that steered us away from older systems, it might well have been compiler issues.

But my thinking is that going out of our way to support Core 2 systems with 32-bit builds indefinitely doesn't make a whole lot of sense, especially since we already went forward with AVX which requires Sandy Bridge at minimum. It probably wouldn't hurt anything to keep a frozen toolchain and SDK for a while, but I am thinking we shouldn't put too much effort into keeping this solution long-term.

[Moonchild]

But my thinking is that going out of our way to support Core 2 systems with 32-bit builds indefinitely doesn't make a whole lot of sense

No, it doesn't, and we can't reasonably do that. We can do so as long as our tree builds with the old toolchain, but after that it's just game over for those systems. As it is, 32-bit is an outlier for what we publish, already, regardless of CPU requirements. There is no "indefinitely" in the most common sense of the word (meaning perpetual). But it is "indefinite" as in not being defined at the moment -- we can support it for as long as we can. And no more.

[moonbat]

We can do so as long as our tree builds with the old toolchain, but after that it's just game over for those systems.

Cue the handwringing and entitled complaining the day that happens

[Moonchild]

Cue the handwringing and entitled complaining the day that happens

And unnecessary attacks against me as a person too, no doubt.

[moonbat]

And unnecessary attacks against me as a person too, no doubt.

Off-topic:
Exactly why I was advocating against touching the Twitter clone Fediverse with a barge pole.

[Pete321]

Thanks for the update , just did auto update