I had extra RAM in my machine. I wanted to see if I could use it to make the system faster. This post takes a look at that.
In this post (as developed in a later post), I focused on speeding up web browsing by use of a RAM disk. That approach led to complications and actually seemed to make things worse. In addition, the attempt to figure out what was going wrong ran into the problem that the built-in Windows memory tracking tools could misstate memory activity and availability in systems with more than 10GB of RAM. There did not appear to be good alternate tools available to watch what was happening with RAM, in real time, as I started or killed various programs or browser tabs. All of these remarks were subject to the caveat that there might be better solutions, if only I could find them or figure them out. For the time being, my working solution was to forget about the RAM drive, increase system RAM as needed, and use browser tweaks to make the cache larger and to put it on another partition. Pending the purchase of additional RAM, those cache-related steps yielded substantial performance increases. It appeared that a RAM disk might still be useful at some point in the future, but I did not seem to need it to speed up browsing at this time.
I seemed to be having two kinds of slowdowns. One was unpredictable. Sometimes the system would get slow, programs would say “Not responding,” I would have big patches of white space on the screen. Sometimes these were clearly due to a resource-intensive program running, especially GoodSync. At other times, it was not so clear. For the time being, I did not intend to tackle these sorts of problems with a RAM disk.
The other kind of slowdown was predictable. Every time I loaded too many tabs in an Internet browser, the browser would slow down, sometimes very much. I was not using Internet Explorer much at this point, and did not notice it too much there. I was using Chrome, but the problem I was having with Chrome was a problem of Flash crashing. I hadn’t seen indications that that was a memory or speed problem per se. But I definitely was having speed problems with Firefox, which I used for all sorts of random tabs, and Opera, which I used especially for opening videos (e.g., YouTube). People said that a RAM disk could significantly improve browser speed. So I decided to start with those.
A search led to a number of webpages suggesting that a RAM disk surely would speed up Firefox. Actually, there seemed to be two ways to use extra RAM to speed up Firefox. One was not to use an actual RAM disk, but instead just to tell Firefox to set up a RAM cache of a certain size. I wasn’t sure whether that would mean that the RAM in question would then be unavailable for other programs, even if I had only a few Firefox tabs open. I also wondered whether this would perpetuate or even worsen Firefox’s memory leaks in ways that would not be possible with a RAM disk. Besides, I wanted to learn how to set up a RAM disk, potentially for a variety of different purposes. So I didn’t pursue that approach.
The questions at hand, then, were (1) how to set up the RAM disk and (2) how to get Firefox and Opera to work with it. Setting up the RAM disk was apparently best done by using a program designed for the purpose. A search suggested that there were a number of freeware RAM disk programs. Raymond (you gotta love the name) tested 12 RAM disk utilities (some free, some not). The field seemed to be changing rapidly, with more programs becoming free and offering larger sizes over the past year or so, and new programs entering the field – such as AMD’s Radeon RAMDisk, arriving too late for Raymond’s review. For the 12 programs tested, Raymond found these 4K read/write times (4K being the default NTFS sector size), in MB/second (F = freeware, at least for smallish RAM disks):
Gillisoft ($40): 1061/751
Primo ($30): 929/706
SuperSpeed ($60): 552/440
Gizmo (F): 90/171
Bond Disc (F): 1352/947
Dataram (F): 218/175
VSuite (F): 154/147
ImDisk (F): 81/75
Gavotte (F): 505/430
SoftPerfect (F): 1027/549
Passmark (F): 80/74
StarWind (F): 231/202
These numbers suggested that, all other things being equal, Bond Disc offered the fastest freeware option – but, Raymond said, all other things were not equal. Bond Disc, he reported, was not easy to use and had a small RAM disk size limit. Rather, Raymond recommended SoftPerfect. That was fine with me. I saw that SoftPerfect offered an open number and size of RAM disks. People gave it a Very Good rating at Softpedia. I downloaded, installed, and launched it.
I went into SoftPerfect > Disk > Add. It said I had only 969 MB free – which was odd, because I had 12GB of RAM on that system, and was presently running only a few browsers and other programs. But I wasn’t actually surprised; it seemed that Windows (or possibly certain piggish programs) always wound up using whatever was available. I shut down almost everything except SoftPerfect – using Start > Run > taskmgr.exe > Applications tab to verify that things really were shut down – and then I hit SoftPerfect’s Disk > Refresh and then Disk > Add menu picks again. It took a couple of tries and some waiting, but eventually it showed almost 6GB free. I created a RAM disk in drive R, of size 4096MB, Mount as Removable, NTFS format. In the Advanced button, I named it to be RAMDISK.
Then I started firing up my various programs again. I checked in Windows Explorer and saw that I did have Local Disk R. There, I went into right-click > Properties > General tab and saw that only 42MB were used, almost 4GB free. I hadn’t seen an option to load this drive image at system startup. Tools > Settings had an option to launch SoftPerfect automatically with Windows, but what would happen to my drive R on reboot? Another SoftPerfect webpage seemed to say that the RAM disk would be persistent if it was associated with an image on my hard drive. But how to create the image? I tried File > Export Disks. Using the name RAMDISK for the export gave me a file called RAMDISK.xml. I rebooted and, what do you know, there was RAMDISK, drive R, all set to go in Windows Explorer.
I had set RAMDISK to be Removable, as just noted, so now I right-clicked in Windows Explorer and looked for the Eject option. It wasn’t there. OK, I went into SoftPerfect > right-click on drive R > Unmount. It struggled but was unable to unmount. What the hell was I supposed to do now? I needed to move RAMDISK.xml to a more permanent home. I killed SoftPerfect, in the sense of clicking on the X on the upper right corner of its window. This only shut down the front end: I could go, now, into the system tray (bottom right corner of the desktop) > right-click on SoftPerfect > Disk Image > Create. This looked like it was going to do the same thing as I had already done. I bailed out and tried again: system tray > right click > Exit > definitely get me out of here. Still failed to dismount. In Windows Explorer, though, I was able to move RAMDISK.xml to a newly created RAMDISK folder. No idea as to what damage this might do, but it was done. Just to make a nice, clean break of it, I went into Process Hacker (but could also have used taskmgr.exe, as above) to make sure SoftPerfect was not still running. Probably should have done that in reverse order but oh, well.
I restarted SoftPerfect and tried to resuscitate RAMDRIVE. It was not responding. Deleting and importing did not help. The best I could get, with various menu picks, was an error message: “Operation failed with status code = C000009A and image status code = Unknown reason.” SoftPerfect was deeply unhappy with me. Starting over, I deleted the old and created a new RAMDISK. This time, I also saved it using Image > Create Image. When it was done, it gave me a message: “Image file created in ‘X:\RAMDISK\RAMDISK.svi.’ You can now mount it by choosing Image > Mount Image.” It still wasn’t able to unmount it, though. (Some of my notes here were sketchy because I was writing them on the computer where I was doing all this installing and rebooting.) In X:\RAMDISK, I had both RAMDISK.svi and RAMDISK.xml. I probably only needed the former, but I wasn’t sure. I did belatedly discover SoftPerfect’s tech support page, including its RAM Disk User’s Manual. So my timing was pretty good there. What could I say? I was watching American Desi and drinking wine while doing this.
Anyway, that part seemed to be done, if half-assed. So now, the next step: getting Firefox and Opera to use that space. Not to slight the manual, but a search had already led to a page explaining how to move not only the Firefox but also the Chrome cache to a RAM disk. So, sure, I went with both, while I was here. First, I went to my RAM drive, R, and created a folder for the cache. I called it BrowserCache. Then I took these steps:
- Chrome: Close Chrome. Find the Chrome shortcut that I used to start the program. (If I used more than one, apparently I would have to go through these steps with one and then copy it over all the others, and remember not to let it get deleted if Chrome updated itself and gave me new shortcuts.) Right-click on the shortcut > Properties > Target > go to the very end of the existing path, after chrome.exe and also after the ending quotation mark, add a space, and then type –disk-cache-dir=”R:\BrowserCache” > OK. I started Chrome. It already had a half-dozen tabs from the last session. It loaded those. It took a while to load those tabs. It created a new folder, R:\BrowserCache\Cache, and stocked that with about 90 files. I closed Chrome and opened it again. It still took more or less as long to load those tabs. So far, I wasn’t sure what I had achieved. I tried loading a bunch of tabs. Normally, this would cause my Chrome installation to slow down, and eventually I would get a “Page Unresponsive” dialog saying, “The following page(s) have become unresponsive. You can wait for them to become responsive or kill them.” This would start to happen when I had around 25-30 tabs open (maybe less, if they had been videos rather than largely text webpages). It did begin to happen in the doctored Chrome, so apparently this RAM disk was not a fix for that. I stopped before I got the “unresponsive” message, but I did get a bar across the top saying, “A plug-in (Shockwave Flash) isn’t responding,” with an option to stop it, which I took. At that point, the slowing ceased, but some previously loaded tabs were blank, as sometimes would happen without the RAM disk, and I had to refresh them to see their contents. These tabs reloaded very slowly, to a point where I wondered whether the RAM disk was making them worse. Eventually, I decided it was one of those situations where one tab was the key: I had to read and dispose of it, and then another would become readable, and so forth. That was the same as before. Also, at some point, I got the usual “Shockwave Flash has crashed” banner notice. So, no, definitely no miracle fix here from the RAM drive. I couldn’t say I was noticing any speed increase, but the 20% figure may have been true in my case.
- Firefox: Type “about:config” in the address bar > Enter > I’ll be careful > right-click anywhere on the list of items > New > String > type this: browser.cache.disk.parent_directory > OK > enter this string value: R:\BrowserCache > close and restart Firefox. Unlike the situation with Chrome, it seemed, I would not have to worry about making copies of a doctored shortcut. On restart, Firefox gave me a white screen: it showed the name of my homepage on the title bar, but otherwise showed nothing. It continued that way for 10-15 minutes. I killed it and tried again. This time, I got nothing. I used Process Hacker (could have used taskmgr.exe) to make sure firefox.exe was dead, and tried again. This time Firefox showed onscreen for a moment and then vanished. Its icon was still in the taskbar. I used Shift-RightClick on that icon and got Firefox maximized onscreen. It hadn’t loaded my previous session. Fortunately, the Session Manager add-on had saved it, so I was able to restore it. Was Firefox faster now? It did seem perkier, but I couldn’t be sure; maybe it was just in a good mood. I would have to give it some time and see how it turned out. The author of one of the webpages cited above said that he got a 20% performance boost. As with Chrome, I wondered especially whether it would make a difference if I had large numbers of tabs loaded. I loaded a bunch of tabs, using links from my (Google) RSS reader. Performance was sluggish, as always was the case when I had many tabs loaded. It got worse as I went on, and the whole computer was slowing down, not just Firefox. But, as usual, Firefox didn’t crash, even with 58 tabs open. At this point, with both Chrome and Firefox using the RAM disk, and between them having about 90 tabs open, Windows Explorer > drive R > Properties indicated that I was using only about 140MB of the 4GB available on the RAM disk, so the slowness should not have been due to swapping out to the hard drive. It did seem that both browsers were using the RAM disk: the most recently created files and folders were from just a moment earlier, when I had been using Firefox. It seemed that the real slowness in these programs was due, not to the cache, but rather to the program files themselves. Meanwhile, at any rate, Firefox was substantially unresponsive. I killed and restarted it. Now it was OK, but not in any sense fast.
- Opera: A webpage (also containing instructions for Internet Explorer) essentially advised this: type opera:config in the Address bar > search for Cache Directory4 > change the address shown there to R:\BrowserCache > Save > OK > close Opera. Upon restarting, after letting Opera update itself (which was probably overdue; I hadn’t used Opera on this machine for months), I opened a bunch of tabs – 45, to be exact. In this case, they were videos, on YouTube and elsewhere. By the time I had started all those tabs, Opera had slowed to a crawl. I took a look at drive R It now contained about 180MB of files, and that would doubtless increase as those videos continued to download. Distractions due to Firefox and Chrome (above) prevented me from dealing more with Opera now; I continued later (below).
- Internet Explorer. While I did not use IE much, it seemed that I might as well record the steps to relocate and enlarge its cache. The relevant settings were at Tools > Internet options > Browsing history > Settings > set the amount of disk space to use and also click on the Move Folder button to relocate.
At this point, Moo0 System Monitor was reporting only 37% CPU usage, but RAM usage of over 7GB, leaving only about 200MB free, and it was reporting first Firefox, and then Opera, as the chief memory loader. The bottleneck, it said, was memory (i.e., not CPU or hard drive). Granted, there were some 130 tabs open in these three browsers, with video thrown in; but when only 200MB of RAM drive cache was being used, how could these programs be burdening 7GB of system RAM?
I had hoped that the RAM drive would make a substantial difference in the functioning of these programs. This had not happened. There seemed to be two possibilities. Either I had not set the RAM drive up properly for caching purposes, or the thing that I was caching was not the primary culprit when running these browsers. I could believe that there might have been more or better things to do, to cache one or more of the three browsers. For example, a MalwareTips webpage said that I had changed the location of my cache, but not the size. Maybe my RAM drive cache needed to be larger? The additional steps were as follows:
- Chrome: In addition to the
--disk-cache-dirswitch shown above, I now added
--disk-cache-size=1073741824in the same place as above (i.e., shortcut Target). I wasn’t sure if that value, a gibibyte (GiB), would be way too much or not enough; I wasn’t sure if the space thus allocated would be rendered unavailable for other programs, or if that –disk-cache-size switch was just expressing a maximum possible value. I started Chrome, using that shortcut, with 21 tabs open. (That is, I had Chrome set to remember and reopen whatever tabs were open when I closed it previously.) I let those tabs finish loading. It took quite a while, and some of them never did entirely finish. (They may have been perpetually updating certain advertising material. Not sure.) Then I closed and restarted Chrome. I figured there was no way that 21 tabs would use up a gigabyte. I hoped that, on restart, those tabs would load much more quickly – which they should have done, if the RAM disk cache was indeed capturing the bulk of the downloaded material. I can tell you that they did not load instantly. They did seem to load more quickly, but it still took four or five minutes to see icons on most tabs, and Shockwave still crashed. I tried on another, similar machine, with 29 tabs, and essentially the same thing happened there.
- Firefox: I searched for a different location within about:config (i.e., not the browser.cache.disk.parent_directory location cited above). This time, the location was
browser.cache.disk.capacity.I right-clicked on that item and chose Modify. I typed the same number as in Chrome: 1073741824. This, I believed, would use up another 1/4 of my 4GB RAM drive. Then, with 20 tabs open, I closed and restarted Firefox. The results were slightly odd. I had unchecked Firefox > Tools > Options > General tab (or in some versions, the Tabs tab) > Startup > Don’t load tabs until selected. So they were supposed to load right away. But they were not doing that. The tab icons appeared almost instantly, which was great. But I had to visit each tab to get its content to load. Still, it loaded almost as soon as I did visit the tabs. It was an entirely different outcome from what I had seen on Chrome. I tried it again, just in case Firefox was in fact loading the content but just hadn’t finished. This time, I waited for one minute after Firefox showed most tab icons. Then I went to individual tabs. This time, the content was there, on every one. I had not paid much attention to what Firefox did without the RAM drive and these tweaks, so now I tried on another, similar computer where I had not made the adjustment. With 28 tabs open, I killed and restarted Firefox there. Same thing: the icons were up right away, but the content was not up right away, nor was it necessarily up within a minute of showing most icons. I thought the difference might be that, this time, I had several videos loading. I got rid of those and tried again with the remaining 25 tabs. But no, the content was definitely not loaded on the tabs I checked, within a minute after most tab icons were visible. So the RAM drive did seem to be making a difference on the first machine.
- Opera: At this point, a couple of days had passed since the Opera-related steps described above, and the system had been restarted. I started Opera and let it load its 45 video tabs. They did not load well: some were playable, some weren’t. And it took a long time. I wondered if they would reload quickly, now that they were back in the RAM drive’s cache. So I killed and restarted Opera. It still seemed to be struggling to load those tabs. Opera, and the system as a whole, was extremely sluggish. Moo0 and Task Manager (taskmgr.exe) agreed: memory was heavily committed. We had only ~150MB free. The delays were due to major disk swapping. It seemed I might just need more RAM, not necessarily in a RAM drive. At this point, due to other obligations, I had to fight my way out of the Opera bog, think about it further, and revisit the problem later. For posterity, the MalwareTips website was advising a slightly different procedure for Opera too, as for the others: (1) Create a separate OperaCache folder on the RAM drive. (2) Type this into the address bar: opera:config#UserPrefs|CacheDirectory4. (3) Enter the path to the OperaCache folder in the CacheDirectory4 settings. (4) Hit “Save” at the bottom of the “User prefs” settings list.
While the foregoing processes were underway, at one point I rebooted overnight. When the system came back up, it gave me this dialog:
SoftPerfect RAM Disk
One or more volumes could not be mounted. Formatting image failed.
A search led to the discovery that nobody else had yet reported this particular error. I was not sure what had happened – in fact, I did not even recall that I had set the system to reboot overnight. There may have been a crash and an automatic restart. For whatever reason, this error was there when I got up the next morning – and along with it, a dysfunctional Firefox. I assumed the dysfunctionality was due to the fact that the place where the Firefox cache was supposed to be was not there. When I clicked on what Windows Explorer was showing as “Removable Disk (R:),” I got an “Insert disk. Please insert a disk into Removable Disk (R:)” dialog. I didn’t want to have to set up the RAM disk manually each time I rebooted, but I thought I might at least do so this one time, just to move the process along. I started the SoftPerfect program. It showed R as the only entry in its “Disks mounted on logon” list. Right-clicking on that entry gave me a grayed-out list: I could not mount or unmount, etc. In fact, everything in the menu was grayed except Help. But in the status bar at the bottom, it said, “1 RAM disks mounted.” I decided to kill SoftPerfect, making sure in taskmgr.exe that it really was dead. Then I restarted it. Now the menu wasn’t grayed. Right-clicking on R suggested that it was already mounted: the Mount option was grayed. But Windows Explorer still showed it as Removable Disk R, and still asked me to insert something. In SoftPerfect, I went to File > Import Disks. It gave me a dialog: “This will dismount and delete all the disks currently configured. Continue?” That wasn’t what I wanted. I tried Image > Mount Image. It took me immediately to the RAMDISK.svi file. But it wouldn’t let me install as R. So, OK, back in the main window, I right-clicked on R and said Unmount > go ahead and force dismount. R vanished from Windows Explorer. Now I tried right-click > Mount R (right there, not in the Image > Mount image option). R was back. Windows Explorer showed it as just Removable Disk (R:) but did not tell me to insert a disk when I clicked on it. Firefox was still nonresponsive, so I killed and restarted it. Now it was responsive. In Windows Explorer, R contained files and folders; and when I killed and restarted Chrome, it added more files and folders, and seemed more responsive than it had been when I first awoke on that confused morning.
So at this point, I had lots of tabs open in Chrome, Firefox, and Opera. Firefox was somewhat slow but largely functional; Chrome less so; Opera not really at all. I wondered whether I just needed more RAM, rather than a RAM disk. Popular opinion seemed to be that 12GB was a lot, and that’s what I had; but the motherboard would accommodate 16GB, and the increment would cost only $22 – not free, but not much, and probably more sensible than investing time and perhaps money in RAM disk software.
To test the idea that it was just a RAM quantity issue, I watched Task Manager as I killed Opera and its 45 video tabs. (Chrome was already shut down.) Task Manager fluctuated, sometimes showing more and sometime less, but not obviously varying much from the ~130MB RAM that it had been showing as available, out of 7677MB total. And with Opera shut down, Firefox was not obviously more responsive. Hmm. This was not persuading me that another 4GB of RAM would make much difference. I tried killing Firefox, with its 38 open tabs at this point. The Processes tab in Task Manager said that, as I knew, Firefox died slowly. I went away from the machine for a while. When I came back, Firefox was dead, available RAM was at 2095MB, and everything else on the system was responsive.
That made me wonder whether the RAM issue was peculiar to Firefox. I wanted to shut down the RAM disk, so that the full 12MB of system RAM would be available, so I could see what would happen if I restarted Opera with its 45 videos (without Firefox or Chrome running). SoftPerfect said it could not unmount the disk because it was in use. LockHunter was apparently not available for entire drives. I tried Windows Explorer > right click > Disconnect. Nothing happened. I went to the system tray > right-click on SoftPerfect > Exit > confirm dismounting. I got an error: “Dismounting volumes on closure failed: Unmounting volume failed.” The Windows Explorer > Disconnect option still did nothing. I restarted SoftPerfect and went into its Tools > Settings > uncheck the “Launch automatically with Windows” option. I went away for 20 minutes, came back, and tried right-clicking on drive R in Windows Explorer again. Disconnect still gave no feedback; but this time, right-click Properties gave me “The properties for this item are not available.” Maybe the disconnect just took a while.
I verified that SoftPerfect was not running. Yet Task Manager still showed only 7677MB of RAM available – not something closer to 12GB. Possibly I would have to reboot to fix that? With only Firefox running (with about 35 tabs), Task Manager said there was 889MB of RAM available. My reading was that Firefox was presently using most but not all available RAM – that when things got to the vicinity of 150MB available, that was probably the minimum that the system would always keep available. With this extra RAM available, tab switching in Firefox was fairly rapid. Possibly a RAM drive would have helped, but not necessarily by much.
I killed Firefox again. Now I had no browsers running. I did still have some other programs running. Available RAM was at 1905MB. I started Opera again. It took just over a minute, given my download speed, for those 45 tabs to drive available RAM down to 447MB. Now, unlike before, all Opera tabs had icons. CPU usage was not especially high, averaging ~20% across eight cores. Since we were well above the ~150MB floor that seemed to signal major system unresponsiveness, I inferred that those Opera plus 45 video tabs were collectively sucking up only about 1.5GB RAM, and I guessed that Opera itself should be responsive. In this, I was disappointed: the contents of various tabs were malformed (e.g., using the wrong fonts, as sometimes happens when pages don’t load correctly) and I was not seeing pages that were actually ready to play videos (e.g., no play button on a YouTube page). I gave it some time. RAM usage did not increase further. The webpage displays did not seem to be improving. Moo0 said the CPU was hardly working, that Opera was the primary burden, and that memory was the bottleneck. That was puzzling. Available RAM dropped to ~410MB, but seemed to stabilize there. I started going down the line, checking each tab in search of the key tab – the one that was actually ready to play which, I had sometimes found, would then free up another one, once I closed it. Checking the tabs seemed to demand more RAM: available RAM dropped to ~315MB, where it seemed to stabilize once again. But then, a half-hour later, available RAM had climbed to ~520MB. Not sure why.
I found and tried to play that so-called key tab. It wouldn’t play. I went to another tab and refreshed the page. It still did not display properly. It then occurred to me that, of course, without the RAM disk, Opera might be confused: I had told it to put its cache there, and now that disk did not exist. Revisiting the steps described above, I went to Opera’s address bar > type opera:configure. Unfortunately, the screen remained blank. I closed all tabs. Still nothing. I killed and restarted Opera. Still got nothing from opera:configure, nor from the alternative shown above: opera:config#UserPrefs|CacheDirectory4. I killed Opera and went into Control Panel > Programs and Features > Opera. There was no Repair option: just Uninstall. I decided to try rebooting instead. That didn’t do it either. I uninstalled and reinstalled Opera. That didn’t change anything. Was that because I didn’t have it delete my user data when uninstalling? Or because I hadn’t rebooted since uninstalling and reinstalling? As advised in one thread, I tried “opera:config#Enable config URL.” That did nothing, so I closed Opera and tried editing C:\Users\Ray\AppData\Roaming\Opera\Opera\operaprefs.ini (where “Ray” was my username, on Windows 7 x64; others might have to search for that file). In operaprefs.ini, I searched for “Enable config URL.” It wasn’t there. So possibly I needed to add it. I added a line reading “Enable config URL=1.” I saved operaprefs.ini and started Opera. This achieved nothing. I killed Opera, removed that line, and closed operaprefs.ini. I couldn’t figure it out.
I completely uninstalled Opera, rebooted, and reinstalled it. Opera:config worked fine now. I had no more urge to redirect its cache to a RAM drive. I did think, however, that I might redirect it to a drive other than C, so that it would not add unnecessary bulk to my system backups. When I got into opera:config, I also thought I might change a few other things. Repeated searches did not lead to a point-by-point tour through these options and, once again, the spirit of experimentation ruled. In the Cache option, I changed Application Cache Quota and SVG Cache Size to 512000 each. I assumed that meant bytes. If it meant MB, I would probably find out soon. In Disk Cache, I changed Buffer Size to 5120, Multimedia Stream Size to 5120, and both Size and Media Cache Size to 10000000. In User Prefs, I set Cache Directory4 and also Operator Cache Directory4 to the desired alternative drive location (X:\Cache\OperaCache). I also changed the Temporarily Download Directory to X:\Cache. After these changes, I restarted Opera and opened 45 video tabs. Five or ten minutes later, available RAM was above 400MB. Almost all tabs had icons, but as before, those tabs were not playable. Task Manager was still showing me as having only 7677MB total RAM, even though Task Manager indicated that SoftPerfect was not running and Windows Explorer was not showing any drive R. I uninstalled SoftPerfect. Without rebooting, Task Manager now reported that I had 2575MB available and 1892MB free.
These experimentations suggested that a RAM drive could cause more problems than it solved, and that the most efficient solution would be just to add more RAM, if needed. As with the steps just described for Opera, I went back and altered the changes I had made to Chrome and Firefox (above). In the case of Chrome, the Properties > Target line now said
"C:\Program Files (x86)\Google\Chrome\Application\chrome.exe" --disk-cache-dir="X:\Cache\ChromeCache" --disk-cache-size=1073741824, and in Firefox, the about:config >
browser.cache.disk.parent_directory entry now pointed to X:\Cache\FirefoxCache. As with Chrome, I left the
browser.cache.disk.capacity value at 1073741824.
Then I rebooted. Windows Explorer confirmed that the new X: caches for the three browsers were all in current use. Oddly, Task Manager still reported only 7677MB total. It wasn’t just Task Manager; Moo0 was reporting more or less the same thing. Hadn’t it been different previously? Available RAM was falling through the floor, but on the positive side Opera now seemed to be in the process of playing videos on each of the first half-dozen tabs I checked — with Firefox and Chrome each running many tabs at the same time. I killed Opera for the moment, just in case the system was about to have a heart attack. A search led to a Microsoft webpage indicating that Task Manager would report only the first 7GB, in some versions of Windows, on systems with 10+ GB. I checked the other computer. No, Moo0 and Task Manager there were both correctly reporting somewhere near the 12GB of RAM on that system. Did I suddenly have a bad memory module on the first machine? Had SoftPerfect caused this problem? The Microsoft webpage said I should use Start > Run > perfmon.exe to track memory usage. I played with it. It wasn’t especially friendly.
For now, I just checked CPU-Z and confirmed 12GB. But I did want a graph to track RAM usage. A search yielded the impression that there were two kinds: those that were focused on processes, like Process Hacker, which I was already using, and those focused on performance reports, which was what I wanted now. One review showed what appeared to be two potential solutions, for my purposes. I downloaded the first, Free Extended Task Manager. From its name and description, I was concerned that it might just reproduce Task Manager’s failings in a more colorful and mis-informative way. I confess, I did not close all other programs before installing it. Despite the warnings, it was not usually necessary to do so. So that may have explained the error message I got when I started it: “ETML Error: -9.” What that meant, I had no idea. I moved on to the other of the two, AnVir Task Manager. It came with some sneaky crapware installation options that prompted me to try to kill the installation midstream. No joy. Like it or not, I got AnVir and whatever goofy bullshit it managed to sneak past me, probably to be discovered later. Now I was looking at a Configuration Wizard that told me I ws going to get just four small buttons at the top right corner of every window. Say what? These would assist me by offering options to keep the window on top, minimize it to the tray, create a floating icon, or roll it up. Very nice, if I had wanted that. I clicked Cancel. So far, so good: no crapware dialogs, no funky four little buttons at top right. Now, could I just please have a graphic view of RAM? The answer, apparently, was no. I was not sure how the guy in the review got a graph out of this thing, but I couldn’t find that option.
I tried a different search. One resulting webpage led me back to Start > Run > taskmgr.exe > Performance tab > Resource Monitor button. This evidently put me back into perfmon.exe, but from a different angle. I wasn’t sure how to get here from Start > Run > perfmon.exe. But, OK, I was here now. But then, I wasn’t free of the Task Manager RAM error: it clarified, erroneously reporting a total of 10240MB installed (instead of 12GB), of which it said there was a total (somehow) of only 7677MB. Whatever. Moving right along. Were there going to be any other options? After a couple of additional false starts, another search led to an About.com webpage listing 15 desktop gadgets, including at least two of relevance for tracking RAM and an eight-core CPU. The ones I gravitated toward were System Control A1 and All CPU Meter. I downloaded and unzipped them. Someone else said that Speed Test (a/k/a Network Meter) was another popular one. Another About.com webpage said I was supposed to be able to just double-click to run them, but this did not happen. Someone recommended Control Panel > Programs and Features > Turn Windows features on or off > Clear the “Windows Gadget Platform” option (if checked, so as to uninstall first) OR in my case, check that Platform option to install it > OK. That worked. Now I had three desktop gadgets for measuring RAM performance. I liked All CPU Meter the best. It was the only one that displayed actual numbers. Unfortunately, as shown here, they were wrong (i.e., 7GB total, again).
So, OK. For now, the situation seemed to be that I could not obtain an accurate, user-friendly impression of RAM usage — although, if I wanted an inaccurate one, I now realized that I could have just used System Information for Windows (SIW), which I already had installed: it produced a nice little graph, comparable to the one in Task Manager.
With that defeat under my belt, I turned back to the issue of RAM and browsers. It seemed that I could not know what was happening about 7GB of RAM, but I could know approximately when something was *not* happening — specifically, when my computer was not functioning. And, in fact, I did know that that had been happening, when I had too many browsers open with too many tabs.
It seemed that my best research strategy, at this point, might be to buy another 4GB of RAM and see what happened. But I also noticed that relocation and resizing of caches for these several browsers seemed to have made a considerable improvement in their speed and responsiveness. Even with 30 or 40 tabs open, playing videos, these browsers were functional and stable.