I was reading some old threads about the origin of Slack64 and see that Pat included the 64bit build because it benchmarked faster. My own experience so far 64bit was a good bit slower on an older system with 2G of RAM, in fact slow enough to be unuseable.
I've ordered some new hardware and was wondering what would be a good way for me to get some #'s to compare, i.e. whats a good benchmark utility I can test with.
This is my first non-laptop in probably 10 years as well as first AMD processor so its somewhat a big change for myself.

This will be a system for the wife and kids to use in the living room so I don't want to run one for a couple months then swap to another.

If it's any use, I went into this. A good cpu test is a kernel build. Use tyhe same config, and source, and run 'time make -j4' or less.

The 64 bit kernel should be able to discard all tweaks and workarounds for older buggy chipsets, and use 64 bit optimizations.

64bit instructions have larger address operands, which will have an impact on the amount of instructions that can be stored in the processors 1st/2nd level cache. Also, more data will need to be transferred to the cpu during instruction fetching, and given that less will fit in the cache there is likely to be a need for more fetches: its kind of a double whammy.

Some early 64bit capable processors likely had caches that were still sized for 32bit operation, reducing the effectiveness of the cache when in 64it mode. I would hope that newer processors will have been given larger caches to better accommodate the larger instructions now that 64bit mode is becoming the norm, but cache size is one of those things that tends to vary from model to model. How much cache is enough, is a harder question to answer though.

Can't say for certain, but if I were to guess, I'd say that this was likely the issue with your older box.

Could be. What I'd like to do with this new box is a quick compare between both. Install 32bit, run benchmark, install 64 bit, run benchmark. My hope is to see which will be better for the new box. I'm not building anything super big and fast http://www.asus.com/Motherboards/AM1MA/ so I'm thinking I could be in that area where the 64bit overhead will cause slowness.

Agreed. I'm running 64 bit stable and -current on two older dual core boxes that have 2 GB of RAM. They run XFCE just fine, however, the boxes use more system resources than when I ran 32 bit Slackware.

Thats why I want to give 64 bit another shot. I've been meaning to build a simple pc for the living room so replace the samsung Bluray that has to be rebooted all the time. So I bought a little case and system board and an AMD Athalon 5350 and 4G of ram. Its not high end stuff so I thought that might make a good 64 bit test and compare to 32bit and see if I crossed over to where 64 bit is faster.

I haven't tested any older 64-bit systems, so I don't know if there is a performance increase or decrease. With newer systems 64-bit will have better performance with most programs. For maximum performance you may want to rebuild the kernel for your processor and tweak it a bit.

I tried moving to 64 bit maybe a year ago. What happens is I use old laptops for most of my systems because I have a family of 4 so if I replace one then then it goes down to the kids. One is my server/lab and a lot of newer OS's are coming out 64 bit only. M$ stuff, slowaris, etc. So I needed to have a 64 bit system for my virtual host. My old laptop was a Dell Latitude D620 with 2G RAM and I have been able to run 3-4 virtual guests at one time. I found a latitude D630 which had a 64 bit capable CPU and 2G ram so I installed Slackware 64 on it and virtual box, etc. I couldn't even run one Slackware guest and the system would start paging and slow to an unuseable crawl. Even without trying virtualbox it was still slow. So I reformatted and went back to Slackware 32 bit and can now multitask again.
So it seems that 2G ram was too little since the rest of the hardware is basically the same other than 64 bit capable CPU/non 64bit capable.
So now my server side is working well I'm finally building a media pc client and wondering if I try 64bit with 4G of ram will it start swapping?

RAM use in 64-bit is increased, so you may want to upgrade to 4 GB especially if you run VMs. You may also want to adjust swapiness.

I'm doing 4G on the new box. Then I want to split my LAMP environment off the lab box since I'm now considering it production. The unfortunate side effect of my lab box being a laptop is RAM is overpriced. I have less than $200 in this new mini itx system so if the performance is decent I might just build another. I'll probably buy a drive for it though, looking at 50-60 for a 120G or so SSD. but still $250 for a system.

Recommending a benchmark without having an exact description of your workload would come down to just guessing.
From your description the APU and 4GB of RAM in that system should be able to handle a 64 bit distro quite fine. This is indeed not a high power CPU, it is competing with Intel's Atom CPUs and does so quite well, though it is no competition at all to Intel's i3 or AMD's FX4000 or A4/6/8/10 series. Some benchmarks, all done on Ubuntu 14.04 64 bit here: http://www.phoronix.com/scan.php?pag...0_ubuntu&num=1

it is very hard to find a machine that is to weak for Slackware, no mater if 32 or 64 bit
you do not need to waster time with benchmarking performance for a living room pc
check what is better compatible with the software you want to run, that is a better time investion

Well I don't know yet what all it will run. We wasted $500 on a Samsung bluray home theater that had nothing but problems so I spent $40 on an extended warranty when Samsung's warranty was about to expire and spent a couple hours on the phone with the extended warranty company documenting all the problems and they agreed to replace it. Well we got a different model which freezes and locks up so I spent the next year trying to get that replaced and finally gave up. The wife and kids know when the Bluray locks up they just need to reboot it, but the wife tries to keep up on a couple shows on netflix and rebooting and waiting for it to spin up the bluray so you can exit the title menu a couple times per half hour show is getting annoying. It locks up and only power off/on will fix it.

I figured I could throw together a simple PC that would work as well. So its basically going to be AlienBOB's Netflix packages, a web browser and maybe flash for youtube, VLC to play/stream videos and music off our server.

I'm wanting to get a rough idea if 64bit will have the issue I saw before or not so I know weather to install Slackware 64 or 32 and go forward from there.

I have a couple of Dell Dimension 8400's that I use as data base servers. They're both 32-bit with 4G RAM, one runs MariaDB, the other runs PostgreSQL, both are Slackware 14.1 stable, configured identically and are fully patched. They run headless and are administered via SSH (so there's no X running on either one).

I also have two Dell Optiplex 780's. They're 64-bit with 8G RAM, Slackware 64-bit 14.1 stable (fully patched) and are configured identically running Xfce.

I did install Slackware 32-bit 14.1 on one of them and did some comparing between the two (it's not fair to compare the Dimensions against the Optiplexs -- they about 10 years old, fast enough for data base servers but not at all the same thing, kind of like comparing a 2004-era Dodge RAM truck to a 2014 one).

I do a lot of geographic map generation. That means heavy math, lots of calculations, not so much disk I/O. The 32-bit Optiplex did a map of Cold War Europe (from the British Isles to the Urals and from North Africa to the Arctic) with shorelines, rivers, boundaries, country names, populated places (text) and topographic information (colors). The result looks like what you see in an atlas. The 32-bit machine generates all that in about five minutes (takes a lot longer to plot it, though), the 64-bit does it in less than three minutes. The code is identical (compiled on the individual platform), the data are the same and the platforms are the same, the same daemons are running on both.

That "longer plot" is the limits of a printer or plotter, got nothing to do with 32- or 64-bit platforms.

The one difference is the RAM, but the job is compute intensive, not memory intensive (I monitor with GKrellM and don't see a lot of memory use but the cores go like mad). Even though there is 8G available, the 32-bit uses 4G (I didn't have extended memory or whatever that is called enabled). The 64-bt will use whatever it needs but, again, I did not see any ramp-up of memory.

Is that a benchmark? Well, no, not in the traditional sense, probably more anecdotal than anything and I don't think any real lessons can be drawn -- just that 64-bit went speed-fast and 32-bit didn't. For what it's worth.

Both the Optiplx boxes are now 64-bit (it was just an experiment), both haul ass at about the same speed (I can't tell the difference).

My only complaint about 64-bit is that Adobe doesn't provide 64-bit Acrobat (and Acrobat Reader); that's Adobe, got nothing whatsoever to do with Slackware or Linux. Okular works pretty OK but it ain't Adobe Reader and it doesn't do things that Reader does. That isn't much of a complaint and I do get around the shortcomings of Okular with Win7 running in VirtualBox when nothing else will do; Reader is installed in Win7 and works just fine in VirtualBox.

I have zero interest in games or Microsoft anything (Office in particular) so I have no need of Multilib (which is a useful package to all reports but I'm just not interested).

My Dimensions do the job I need them to do: they sit in a closet mumbling to themselves and wake up and go when called upon. My Optiplexes do the jobs I need them to do. The world has gone to 64-bit and it ain't gonna back but there's no reason to throw out a working 32-bit box either. In my experience, 64-bit machines are faster, better, cleanre and I'm going to stick with them.

Hope this helps some.