Windows Accelerators - Do They Really Work?

danila asks: "Today I came across an intriguing review of Windows tweakers on a Russian technology news site. Among the plethora of traditional registry tweakers, the review mentioned Hare 1.5.1. The developers promised nothing less than up to 300% speed increase, 10% FPS increase in 3D games, automatic RAM preservation and even a wizard that automatically cleans and optimizes Windows. It also had AntiCrash 3.6.1 a program to prevent up to 95.8% of Windows crashes. Understandably, I was both intrigued and suspicious since it sounded too good to be true." Has anyone tried this piece of software with any degree of success? How successful are other "windows accelerators" at improving Windows performance? "After a little research I found that download.com didn't have it and there are precious few reviews of this revolutionary software online, but that it was endorsed by McAfee and that developers touted conformance with Microsoft's interface guidelines as an important feature.

Still suspicious, I gathered all my courage and installed both programs (silently preparing for something like Bonsi Buddy or XXX Toolbar) on my Win2k Pro machine (P4 1.6/512Mb). Truth be told, after several minutes I was blown away. Obviously I can't tell how well every promised features works, but disk caching (and pre-fetching) that Hare does is outstanding and display performance improved enough to scare me - windows were opening, minimizing and redrawing without the delay I was accustomed to.

The question is -- is it real or was I fooled by some clever placebo tricks? And if it is real, why isn't the Web full of success stories involving Hare and AntiCrash? Why isn't everyone installing them on every Windows machine in the world? And a rhetorical question -- why doesn't Microsoft incorporate some of the features into its operating systems."

Comments

Win4Lin

[Anonymous Coward]

gave me the ability to boot Windows98 in 7 sec. and provided me with the most stable Windowsxx ever!

Old software...

[Skates1616]

Both of these programs had their last revisions in late 2002, so it remains to be seen how effective they are now, or this is just some marketing BS...

Hare

[SynKKnyS]

I tried Hare and it never seemed to make a difference at all. It did have many interesting options, though.

The only program that ever seemed to speed anything up was O&O Defrag (oo-software.com) who have a background defragger. Leave your computer, and the defrag turns on. When you come back, it is off in anywhere from instantly to a minute. The program also has a nice complete defragger to boot.

It's possible, I suppose.

[Anonymous Coward]

Although doubtful, I wouldn't say such software is impossible. Sometimes some rather neat hacks can be pulled. Example: The Apple Macintosh IIsi came with 1 Mb of on-board memory. This memory was very slow, AND it was shared with video. If you installed SIMMS, however, this memory could actually be operated at a faster speed (70ns max if I remember correctly) than the onboard memory.

Some hacker wrote a program called IIsi RAM Muncher which allocates the first megabyte of memory on start-up, and then does nothing with it. Result? All your stuff runs in the faster SIMM memory. The speed increase could be as much as 400% - not bad for giving up 1 meg of RAM.

Re:It's possible, I suppose.

[runderwo]

Many older PC memory controllers don't cache more than 64MB of system RAM. In the case that you want more than 64MB of RAM, the net result will usually be a slowdown in overall performance due to the kernel running out of uncached memory (it is loaded at the top of memory).

The solution is the slram MTD driver. Pass mem=64M to the kernel to force it to load within the cached region, and then use slram to allocate the rest of system memory. This gives you a block device which you can use as a scratchpad, tmpfs, or even better, a swap area! Swapping to RAM (even uncached RAM) is much faster than swapping to disk in any case. I did this on my Toshiba Tecra 500CDT. The system has 144MB RAM but only 64MB is cacheable. It runs like a champ, and I don't need a swap partition on disk this way.

The first time I described this idea to someone, he thought I was crazy. I realized why the reaction was such. Just think if you told someone that the magic secret to speeding up your system is to "put your swapfile on a RAM disk"...

Connectix Ram Doubler and CrashGuard

[sublimespot]

Connectix Ram Doubler and CrashGuard worked beautifully on Mac back in the day. I always wondered if the same thing could be done on PC as well as Connectix did it for Mac.

Re:Connectix Ram Doubler and CrashGuard

[jruschme]

Ironically, Connectix did briefly make a version of RanDoubler for the PC. IIRC, it was very buggy and quickly killed off.

From its description, Hare appears to combine the functionality of RamDoubler and other utility called MemOptimizer. Similarly, the same company makes a utility called Zoom which sounds a lot like the old StartupDoubler utility for the Mac.

>>

Re:Connectix Ram Doubler and CrashGuard

[redJag]

My initial reply was going to be along the lines of "Ram Doubler" was just an early implementation of virtual memory, but that was wrong..

Just an FYI for the non-Mac user, the classic Mac OS would allocate a given amount of RAM to each application at launch, whether it used it or not was up to the application. Ram Doubler would reclaim unused, allocated memory and give it to other applications that needed it. It would also compress memory if it hadn't been used for a while, and as a last resort would page out to the harddrive (like the Finder would already do).

Re:Connectix Ram Doubler and CrashGuard

[madsenj37]

Ram Doubler and Speed Doubler worked beautifully but became unnecessary after OS 9 was released. Surf Doubler worked well also. I never tried crashgaurd. Once OS 9 came out with more PPC code, Connectix products for speed became obsolete.

Window Drawing

[clockworx]

Could be they just turn off lots of the built-in delays that MS has in the system. You can turn of Window animation, menu item slide-in/fade-in/fade-out, and turn the delay to 0 for opening submenus. I do this with every install, and users always think the system is amazingly faster :)

Interesting

[sublimusasterisk]

From the Hare website faq:

* Hare technology: the core of Hare is a re-written Kernel, working at up to 88-bit (instead of the standard 32-bit) and accelerating most basic system actions by acting as the Windows Kernel. This is done by triple-buffering all I/O data, in order to achieve an emulated 88-bit Kernel. This technology is fully safe and we have implemented safeguards in order to make it impossible to damage your computer.

That seems a bit suspicious. 88-bit!? Ok, so it's emulated. That still seems like 1) a strange number (not 64, not 128) and 2) would "emulated" 88-bit architecture really work? Isn't the CPU's inherent 32-bitness (or 64-bitness) the end-all anyway?

* CPU Tasking: the CPU Tasking technology's goal is to give more CPU to the program you currently use. Even if you don't know it, there are a lot of programs working in background and sucking CPU from your frontmost application - the CPU Tasking will know how much CPU you must give to each application."

Doesn't Windows already do this?

Re:Interesting

[Anonymous Coward]

Like the first AC said, but with different words, this 88-bit reference almost certainly refers to using the native floating-point registers.

Before IEEE came along and standardize floating point math operations, every ISA vendor had their own way of doing floating point (some even had more than one way). Intel systems now support the IEEE format, but I'm pretty sure they kept their native 88-bit (instead of the now common 64-bit "double precision") format around too.

They are probably using the floating point registers to either do some extra low-precision integer math in parallel or just as scratch space instead of the stack and/or L1 cache. In the past, access to the FP registers sucked worse than hiting cache, but maybe Intel/AMD tweaked things up a bit with recent generations to improve that.

Re:There is a simple reason

[cuzality]

there is no way any windows desktop can beat that speed.

Don't be too sure.

Lately I've been using LiteStep, a Windows version of the Unix window manager AfterStep, and I have to say I have been very impressed with the overall improvement in performance. I've got an old Celeron 800Mhz notebook with 256MB of RAM that was struggling under standard WinXP Pro, even with all window-dressing (so to speak) turned off (like zooming windows, big desktop background graphics, etc.). This was especially obvious when I would use a removable wireless adapter card -- Firefox was sluggish and even unresponsive at times. (And seriously, this was a completely stripped-down environment -- no extraneous services running or background programs sucking up available resources.)

But since switching from Explorer to LiteStep as my default shell, just about everything about how Windows works has improved in terms of responsiveness and speed in general. My frustration level has been seriously cut down. And on top of that, my wife now refuses to use the laptop because of the new shell -- what a shame.

I'd bet a WindowsXP machine using LiteStep as the shell could keep up with just about any stripped down window manager for Linux like Fluxbox.

Re:A long time ago...

[Zerbey]

I had a similar piece of software that claimed to turn my 486SX into a 486DX with a "3-fold" increase in speed.

It didn't work of course, but it did allow me to run some software that insisted on an FPU (Quake 1 was... interesting :-)). My raytracing software actually ran noticeably slower with it as it tried to make use of the emulated FPU :)

Increasing framerates

[Bob McCown]

Im trying to get my framerate up to where I can even play games. Ive got Win2k, a Radeon 9600, 1.2 processor, gig of ram. I cant even get 5fps out of my box, and have no clue what is causing it. Are there utilities around showing what drivers are loaded yet unneeded? HELP!

Not worth it.

[MisterFancypants]

Some of these products do actually make a performance difference, usually by altering the way the system does memory management. However, the performance difference is usually usage dependent so you may or may not see the difference based on what you do with your system.

Having said that, in my experience these programs virtually all cause some instability or other that makes them just not worth it. I wouldn't run one of these for the same reason I don't overclock my systems -- the couple of percentage points of increased performance just isn't worth the increased risk that my system might die at some critical moment, causing me to lose hours or more of work.

YMMV.

Another Simple Reason: APPARENT speed gains

[Anonymous Coward]

In creating GUIs for programs I've worked on, I've noticed that people will THINK I've made something a lot faster if only I tweak how the slow task looks on the screen. For example, let's say my program parses an XML file in 5 seconds. I have some options:

1. Let it freeze my GUI
2. Change the cursor to the 'busy' cursor
3. Show the user a progress bar

Of course the easiest to do is option 1, but to users this also appears to be the slowest. 2 is an improvement -- but still seems kinda slow. Users think option 3 is blazingly fast for some reason -- and EVEN BETTER is if you create a progress bar that fills up to 100% multiple times before it's done (users no doubt think "WOW, look at that progress bar go!").

But back to the point: windows accelerators. I remember finding a registry tweak a LONG time ago which eliminated the short delay between displaying 'trees' in the start menu. Whenever ANYBODY used my computer (while this tweak was in effect), they always told me how fast it seemed to them. Was it faster? Well, yes, a 0.1 second delay was removed, but really it didn't make what you were trying to do go any quicker.

I guess my point is that speed doesn't matter so much as appearance.

Re:There is a simple reason

[Stevyn]

I agreed with you about point #2 when I used Mandrake. It was slower than windows. I was using KDE because you know what? I like a full fledged window manager. I shouldn't have to settle for IceWM just to get the speed of windows xp. Anyway, I switched to gentoo and it's faster than windows. I don't know if it's because I compiled most of the software (I did a stage 3 and compiled from there) or because they structure it better, but it's a hell of a lot faster than mandrake.

W2K for gaming...

[Anonymous Coward]

I've got a pair of otherwise identical boxes, one with W2KSP4 and the other with XP Pro SP1. They are:

ABIT NF7-S mobos, Athlon 3200+ cpu's, Saphire Radeon 9500NP video cards softmodded and overclocked to 9700 specs, Mushkin Level One PC-3500 memory 2x256MB in dual channel mode, Logitech keyboards and MX-510 optical mouse.

In both UT2003 and UT2004, the FPS with everything turned on max highest details, and resolution at 1024x678, the W2K box is about 5 FPS faster on average than the XP box. Other than seeing the FPS display slightly different numbers on the screen while playing, I cannot tell the difference bewteen them. The gaming performance is virtually identical on both operating systems.

Re:There is a simple reason

[MikeBabcock]

There are commercial graphics accelerators for Windows that work very well -- commercially developped PCI and AGP video drivers that are more well-written than the stock ones from the manufacturer. See SciTech for more info ... and no, I don't work for them :)

PS, they do the same thing for Linux XFree86 drivers as well.

My input

[NeoSkandranon]

I've seen older (win9x and win2k) installs fragged by "registry cleaners" who swear up and down some key is unneeded or incorrect and then lo and behold, you get rid of about a dozen of those...and stuff starts working funny. Since then I steer clear of any kind of "Clean up your computer" stuff. The only apps I use in that vein are Spybot, Spysweeper and Norton AV

Doubting Thomas

[ThisIsFred]

I seriously doubt there is anything revolutionary here, and most likely you're trading one thing for another. If I disable XP system restore and file indexing, I get better load times, but I don't have the capability to restore the previous configuration, and my searches take longer. I don't care about those, so I disable them, and it's a win for me. But I thought one of the improvements of NT-family desktop operating systems was not allowing UI stuff to hog so much processor time. Sounds like a step backwards to me. And higher framerates aren't everything. I'd rather trade 10 frames if it means I'm not losing client update packets to choke, or that my keyboard input isn't being ignored.

Stop it with the grandma

[SmallFurryCreature]

It is insulting, some grandmothers broke bloody enigma, worked on the first computers, developed the first computer languages. If you don't get linux or for that matter windows just admit it.

You would be suprised how many grandmothers worked in business and had Unix come in as the newbie. To them linux will be childs play just as soon as someone actually allows them to get their hands on it and the teenage looser grandson doesn't think he knows best.

For the rest I agree with you. My linux desktop been more then ready. I code, surf, watch movies all a lot easier and faster then on a windows machine. I still can't understand all the stuff about codecs. Movies just work for me. Got to love mplayer. Linux not ready? Windows is not ready. Windows got the codecs, just not the architecture to install them all easily.

Re:There is a simple reason

[Proc6]

Or Hyperthreading. Or like me, both :) 4 CPUs in the Task Manager is cool even if 2 are fake :P

Re:There is a simple reason

[Otter]

Why is it that everyone tosses around "not ready for the desktop" as if it's a meaningful phrase?

As the use of capital letters in my part was meant to suggest, I wasn't referring to any use of Linux as a desktop OS (I'm posting this from a Gentoo box running WindowMaker) but to the ongoing UI squabbling that typically falls under the banner of Linux Is Ready For The Desktop.

Same reason why taking a "long" short cut in a car

[anti-NAT]

works.

I find avoiding sitting at traffic lights on major roads by going via the back streets, even if the trip takes longer in both time and distance terms, feels shorter. I'm keeping moving, so I feel like I'm getting somewhere for more of the trip.

Which tends to indicate that if you can distract a human mind from making "time monitoring" the current focus, a human mind will not perceive lengths of time as accurately.

I think your progress bars are having the same effect as me taking the backstreets.

This and Doom3

[_KiTA_]

I installed Hare and enabled the various video options. Then I went into Doom 3. Holy crap.

It made a huge difference. No more jittery lag when mobs appear, no graphics lag when waving the camera around too fast, etc.

The only problem is that I suspect the game is running TOO fast now. The chainsaw "feels" faster to rev up and animate. Going down an elevator I could have sworn took a while last time only took a few seconds this time, etc.

I suspect that Hare runs on the same principle as GEAR, the old MMORPG eploit. Basically what Gear would do is force a program, like, UO, EQ, AO, etc, to run faster than it should by speeding up the clock timer. The games, which had delays and waits built in to slow the player down, suddenly didn't have to wait, and ran a ton faster.

I also highly suspect that Hare automatically puts the current active program on the highest or second highest priority. This would also explain why Doom3 gets so much more FPS.

I'm gonna go play more doom 3 as.. er, research, ya, research to test this theory out. I suspect that playing during a cut-scene is going to be interesting.