<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	
	>
<channel>
	<title>
	Comments on: More PS/2 upgrades!  More RAM, More advanced SCSI!	</title>
	<atom:link href="https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/feed/" rel="self" type="application/rss+xml" />
	<link>https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/</link>
	<description>Fun with Virtualization</description>
	<lastBuildDate>Sat, 30 Sep 2023 21:03:21 +0000</lastBuildDate>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	<generator>https://wordpress.org/?v=7.0</generator>
	<item>
		<title>
		By: Chris M.		</title>
		<link>https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339661</link>

		<dc:creator><![CDATA[Chris M.]]></dc:creator>
		<pubDate>Sat, 30 Sep 2023 21:03:21 +0000</pubDate>
		<guid isPermaLink="false">https://virtuallyfun.com/?p=13243#comment-339661</guid>

					<description><![CDATA[In reply to &lt;a href=&quot;https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339634&quot;&gt;neozeed&lt;/a&gt;.

That or find an AOX MicroMaster 386 or the later Kingston MCMaster 386/486. Both are CPU upgrade cards for the Model 50/60 and have additional RAM slots. I have a AOX 386 card here with a missing 386 and broken SIMM retention tabs. Stupid plastic garbage.]]></description>
			<content:encoded><![CDATA[<p>In reply to <a href="https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339634">neozeed</a>.</p>
<p>That or find an AOX MicroMaster 386 or the later Kingston MCMaster 386/486. Both are CPU upgrade cards for the Model 50/60 and have additional RAM slots. I have a AOX 386 card here with a missing 386 and broken SIMM retention tabs. Stupid plastic garbage.</p>
]]></content:encoded>
		
			</item>
		<item>
		<title>
		By: Malcolm		</title>
		<link>https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339651</link>

		<dc:creator><![CDATA[Malcolm]]></dc:creator>
		<pubDate>Sat, 30 Sep 2023 18:41:16 +0000</pubDate>
		<guid isPermaLink="false">https://virtuallyfun.com/?p=13243#comment-339651</guid>

					<description><![CDATA[This one post managed to revive so much trauma.

I hadn&#039;t noticed how bad the conventional memory situation is in OS/2 1.x before - it didn&#039;t include MEM.EXE and was such a bad runtime for DOS programs it never mattered.  Looking now I think you nailed it with bimodal drivers - I hadn&#039;t heard of this before, and while it makes sense to have a single mouse driver for DOS and OS/2, a lot of drivers look to be loaded here that shouldn&#039;t be.  I&#039;m seeing things like PMDD for example - what is DOS supposed to do with that?

For the leaked XMS, my understanding is that HIMEM.SYS is providing mechanics to copy memory around and also providing a simple allocator to track which XMS regions are in use and provide handles to them.  Unfortunately this is all before the current concept of a &quot;process&quot; so there&#039;s no reliable way to clean these up on exit.  I suspect you&#039;re seeing leaks from EMM286 rather than Windows, because a) it needs to allocate XMS and b) it sounds less mainstream/thoroughly tested.  I&#039;d be curious if you use 1Mb EMS instead to see if the leak still happens (3Mb being much larger than a 286 would typically do.)  Oh, and agree with Stefan that it needs to copy as opposed to remap, which would make it very slow.

For EMS on OS/2 1.x, MSD is showing the memory range needed for it is available, and the &quot;bimodal driver&quot; design would actually help here.  It seems like it ought to be possible to do.  Although rather than do that, it might be more interesting to turn OS/2 1.x into a DPMI server...]]></description>
			<content:encoded><![CDATA[<p>This one post managed to revive so much trauma.</p>
<p>I hadn&#8217;t noticed how bad the conventional memory situation is in OS/2 1.x before &#8211; it didn&#8217;t include MEM.EXE and was such a bad runtime for DOS programs it never mattered.  Looking now I think you nailed it with bimodal drivers &#8211; I hadn&#8217;t heard of this before, and while it makes sense to have a single mouse driver for DOS and OS/2, a lot of drivers look to be loaded here that shouldn&#8217;t be.  I&#8217;m seeing things like PMDD for example &#8211; what is DOS supposed to do with that?</p>
<p>For the leaked XMS, my understanding is that HIMEM.SYS is providing mechanics to copy memory around and also providing a simple allocator to track which XMS regions are in use and provide handles to them.  Unfortunately this is all before the current concept of a &#8220;process&#8221; so there&#8217;s no reliable way to clean these up on exit.  I suspect you&#8217;re seeing leaks from EMM286 rather than Windows, because a) it needs to allocate XMS and b) it sounds less mainstream/thoroughly tested.  I&#8217;d be curious if you use 1Mb EMS instead to see if the leak still happens (3Mb being much larger than a 286 would typically do.)  Oh, and agree with Stefan that it needs to copy as opposed to remap, which would make it very slow.</p>
<p>For EMS on OS/2 1.x, MSD is showing the memory range needed for it is available, and the &#8220;bimodal driver&#8221; design would actually help here.  It seems like it ought to be possible to do.  Although rather than do that, it might be more interesting to turn OS/2 1.x into a DPMI server&#8230;</p>
]]></content:encoded>
		
			</item>
		<item>
		<title>
		By: neozeed		</title>
		<link>https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339634</link>

		<dc:creator><![CDATA[neozeed]]></dc:creator>
		<pubDate>Sat, 30 Sep 2023 10:37:09 +0000</pubDate>
		<guid isPermaLink="false">https://virtuallyfun.com/?p=13243#comment-339634</guid>

					<description><![CDATA[In reply to &lt;a href=&quot;https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339622&quot;&gt;Zir Blazer&lt;/a&gt;.

Ive been under the impression that LOADALL is much more faster, but yes there is this implied loss of speed from A20 gating, changing modes, and shuffling data around.  I didn&#039;t think the overhead would be so bad, when looking at disk speed timings when comparing a few himem.sys types (xmsmmgr vs Windows 3.1 himem.sys) and smartdrv on / off, which was so minimal.

Indeed the only way to tell is a legit EMS card, however at the moment the only MCA card market I&#039;ve managed to crash is the SCSI market.  RAM cards are still ~ £100-£250 range.  Which is a shame.  I should have probably jumped on an 8580 20Mhz motherboard for £130, that would have been a big bump in performance, and of course giving me access to v86 mode &amp; 32bit software.]]></description>
			<content:encoded><![CDATA[<p>In reply to <a href="https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339622">Zir Blazer</a>.</p>
<p>Ive been under the impression that LOADALL is much more faster, but yes there is this implied loss of speed from A20 gating, changing modes, and shuffling data around.  I didn&#8217;t think the overhead would be so bad, when looking at disk speed timings when comparing a few himem.sys types (xmsmmgr vs Windows 3.1 himem.sys) and smartdrv on / off, which was so minimal.</p>
<p>Indeed the only way to tell is a legit EMS card, however at the moment the only MCA card market I&#8217;ve managed to crash is the SCSI market.  RAM cards are still ~ £100-£250 range.  Which is a shame.  I should have probably jumped on an 8580 20Mhz motherboard for £130, that would have been a big bump in performance, and of course giving me access to v86 mode &#038; 32bit software.</p>
]]></content:encoded>
		
			</item>
		<item>
		<title>
		By: Stefan Ring		</title>
		<link>https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339628</link>

		<dc:creator><![CDATA[Stefan Ring]]></dc:creator>
		<pubDate>Sat, 30 Sep 2023 08:39:22 +0000</pubDate>
		<guid isPermaLink="false">https://virtuallyfun.com/?p=13243#comment-339628</guid>

					<description><![CDATA[In reply to &lt;a href=&quot;https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339622&quot;&gt;Zir Blazer&lt;/a&gt;.

The way I imagine EMM286 works is by copying to and from XMS memory every time the EMS client changes the mapping. At least I cannot imagine how it would work in any other way without a V86 mode (and without support from the chipset). Yes, it needs to switch modes, but it also needs to copy memory around a lot, which likely plays a much larger role.]]></description>
			<content:encoded><![CDATA[<p>In reply to <a href="https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339622">Zir Blazer</a>.</p>
<p>The way I imagine EMM286 works is by copying to and from XMS memory every time the EMS client changes the mapping. At least I cannot imagine how it would work in any other way without a V86 mode (and without support from the chipset). Yes, it needs to switch modes, but it also needs to copy memory around a lot, which likely plays a much larger role.</p>
]]></content:encoded>
		
			</item>
		<item>
		<title>
		By: Zir Blazer		</title>
		<link>https://virtuallyfun.com/2023/09/29/more-ps-2-upgrades-more-ram-more-advanced-scsi/comment-page-1/#comment-339622</link>

		<dc:creator><![CDATA[Zir Blazer]]></dc:creator>
		<pubDate>Sat, 30 Sep 2023 06:38:02 +0000</pubDate>
		<guid isPermaLink="false">https://virtuallyfun.com/?p=13243#comment-339622</guid>

					<description><![CDATA[In your case, EMS being slower than XMS makes perfect sense because you&#039;re pretty much emulating it on top of XMS via Software. So using EMS actually incurs all the penalties from XMS 80286 Protected Mode/LOADALL shenanigans plus EMS Bank Switching emulation on top of that. I suppose than it would have been a great solution for people so that they can make use of EMS without getting a physical EMS card for Software without XMS support.
Of course, now the question remains of how it would perform with a true MicroChannel EMS card...]]></description>
			<content:encoded><![CDATA[<p>In your case, EMS being slower than XMS makes perfect sense because you&#8217;re pretty much emulating it on top of XMS via Software. So using EMS actually incurs all the penalties from XMS 80286 Protected Mode/LOADALL shenanigans plus EMS Bank Switching emulation on top of that. I suppose than it would have been a great solution for people so that they can make use of EMS without getting a physical EMS card for Software without XMS support.<br />
Of course, now the question remains of how it would perform with a true MicroChannel EMS card&#8230;</p>
]]></content:encoded>
		
			</item>
	</channel>
</rss>
