Realbench v2 [Discussion Thread/Download Links]

[MatsGlobetrotter]

Hmm WillyK,

To run any of the Benchmarks you find on HWbot would yield exactly the same result in terms of outdated services and then potentially a security threat. I have had to install old Mediaplayers, media encoders just to be able to run a benchmarker software. If you are afraid of the security loopholes to that extent then its simply not possible to use any benchmark software on your production machine. Every single change in a package will change the scores uncontrollably and thus not be comparable. claiming it has little effect is wrong. otherwise why is everyone turning of even a simple service in the taskmanager even if it is 5kb. large.. simply because it takes space in the processor slowing down the time for the benchmark.

I would offer two solutions to your dilemma.

1. install the realbench only when you intend to run it meanwhile turn off internet. Then de-install it fully afterwards. You seem to have good software to do the cleanup and checks so the disk should be clean afterwards.
2. Get a separate HDD or SSD and install Win7 for example only to run the benchmark softwares that then is completely isolated from your normal production system.

I also run some benchmark software on my production machine but that wont give you neither good results nor good safety for your normal files simply due to the potential crashes if you overclock and run into multiple crashes which are common. in that scenario external viruses would be the least of your problems. Even the smallest 64GB HDD or SSD at around 50 - 70 USD would probably do the trick unless you have an old one lying around.

That would, in addition to get you more security also make you more aware of the bloatware that might slow down you machine in general (secure or insecure) with the inclusion of automatically detecting version updaters which are serious culprits in slowing down machines. Learned this by testing myself.

Regards,

[twisted1]

Hmm WillyK,

To run any of the Benchmarks you find on HWbot would yield exactly the same result in terms of outdated services and then potentially a security threat. I have had to install old Mediaplayers, media encoders just to be able to run a benchmarker software. If you are afraid of the security loopholes to that extent then its simply not possible to use any benchmark software on your production machine. Every single change in a package will change the scores uncontrollably and thus not be comparable. claiming it has little effect is wrong. otherwise why is everyone turning of even a simple service in the taskmanager even if it is 5kb. large.. simply because it takes space in the processor slowing down the time for the benchmark.

I would offer two solutions to your dilemma.

1. install the realbench only when you intend to run it meanwhile turn off internet. Then de-install it fully afterwards. You seem to have good software to do the cleanup and checks so the disk should be clean afterwards.
2. Get a separate HDD or SSD and install Win7 for example only to run the benchmark softwares that then is completely isolated from your normal production system.

I also run some benchmark software on my production machine but that wont give you neither good results nor good safety for your normal files simply due to the potential crashes if you overclock and run into multiple crashes which are common. in that scenario external viruses would be the least of your problems. Even the smallest 64GB HDD or SSD at around 50 - 70 USD would probably do the trick unless you have an old one lying around.

That would, in addition to get you more security also make you more aware of the bloatware that might slow down you machine in general (secure or insecure) with the inclusion of automatically detecting version updaters which are serious culprits in slowing down machines. Learned this by testing myself.

Regards,

You're absolutely correct Mats, updating benchmarks that work and score evenly is a no no.

Internet when benching? No go, that should be disabled from BIOS

[jab383]

Mats touched on two separate problems with the same solution. First the security of older software used for consistent, comparable benchmarks. They are vulnerable - no argument there.

The second is OS damage from overclocking. I dialed in a 4790K followed soon by a G3258 Pentium. The 4790K work was during a competition where there was much to learn. Much tweakage and crashing. Had to redo RAM timing with the Pentium since its IMC wouldn't reach 2400. In all I wrecked Windows about twice a week in that process. I did all that in the second option of a dual boot installation and I have a backup images. Recovery takes about 5 minutes including getting the backup HDD and plugging it in.

My 24/7 OS is loaded with security, nice-to-have services and gets updated regularly - manually when I have the time, nothing automatic.

The point is that risks are there and they are real. They are also worse when overclocking and benching. I don't want to compromise the benchmarks by bloating the OS nor by 'cheating' with the latest software updates. Backup image of the OS is a good solution to both risks for me.

Jeff

[Myk SilentShadow]

Hold on just a cotton pickin minute, you mean to tell me that updating any component of a benchmark is considered cheating? since when? and how the hell is it we're only just learning this now, after 2 years of Vizman's Challenges?

[twisted1]

Mats touched on two separate problems with the same solution. First the security of older software used for consistent, comparable benchmarks. They are vulnerable - no argument there.

The second is OS damage from overclocking. I dialed in a 4790K followed soon by a G3258 Pentium. The 4790K work was during a competition where there was much to learn. Much tweakage and crashing. Had to redo RAM timing with the Pentium since its IMC wouldn't reach 2400. In all I wrecked Windows about twice a week in that process. I did all that in the second option of a dual boot installation and I have a backup images. Recovery takes about 5 minutes including getting the backup HDD and plugging it in.

My 24/7 OS is loaded with security, nice-to-have services and gets updated regularly - manually when I have the time, nothing automatic.

The point is that risks are there and they are real. They are also worse when overclocking and benching. I don't want to compromise the benchmarks by bloating the OS nor by 'cheating' with the latest software updates. Backup image of the OS is a good solution to both risks for me.

Jeff

I seriously need to make image backups of my bench OS's too. It's getting really old reinstalling windows all the time. The more BSOD/Crashes you have on an OS the more unstable it gets. X79 seems more sensitive to this, a perfectly stable OC profile can be unstable as hell the next day. And X79 works in mysterious ways, SB/Ivy/Haswell is a walk in the park to OC compared to SB-E

[WillyK]

I believe that as HW/SW evolves, the RB would need to be upgraded anyhow in order to stay capable of serving its purpose properly (since it's not based on some "timeless" design/code). The only question is when. Whenever it may happen, it'll trigger all the above mentioned issues regardless. So from this viewpoint it makes no difference if any RB update happens tomorrow or next year. (Actually, the sooner, the better.)
It might be wise to consider how others are maintaining their benchmarking gear (e.g. Futuremark).

To me RB is mainly useful as a real-life stability test. There already exist a myriad of widely recognized benchmarking tools out there so not sure we need yet another one. The RB's real value is precisely its real-life stress testing capability, as opposed to the synthetic tests most of the other tools are offering. But ... that's just my opinion

[twisted1]

WillyK: RB is on hwbot now And we(at least me) can't get enough of competitive benchmarks

[jab383]

Hold on just a cotton pickin minute, you mean to tell me that updating any component of a benchmark is considered cheating? since when? and how the hell is it we're only just learning this now, after 2 years of Vizman's Challenges?

That's why 'cheat' is in quotes. Some count - Realbench checks - and some don't and some require update.

Jeff

[MatsGlobetrotter]

Dear WillyK,

I am referring to exactly Futuremark on my comments. We all love running 3dMark 2001 as it is representing different era in how graphics was processed in the computers. Futuremark is releasing new benchmarking tools all the time as the technologies change and people use the new ones as well as the older ones. (reminds me i still need to get a key to use PCmark 04) Even on the simple futuremark system tool you cant run the latest version with for example PCMark05, you will never get a result if you do. Thus they only last what 3 years in one version anyway.

If we need or not another tool is up to each and everyone to consider and use as they wish, I mean even as a stress tester there has been numerous ones already, there is nothing unique with Realbench in that sense. They all test and stress different areas of a computer. Infact i think it overstresses the computer as seemingly I can run anything else but Realbench on some speeds and with that I also mean my 3d live rendering architectural softwares which are a killer for both CPU, GPU as well as HDD.

Im using most softwares to test and stress various scenarios as I like seeing how my computer fare among other computers. Even my old crappy laptops sometimes do good :-)

Some test benches like Hwbot Prime allow you to use any Java version to run for example.. Well sure dont use the ones from the last 4 years as they are all bloated and slow so you have to dig up an old slim one.. Took me a week to find what to trust among all the information on internet. There was over 30% difference in the result and then i have not even tried the 32 bit version yet....

Regards,

[DisturbedJim83]

Thought I'd add my results though can't get the screenshot to upload

system specs are:

CPU: i5 4670K@4.4ghz
MB:Asus Maximus VI Formula
RAM: 16GB(2x8GB)Corsair Vengeance Pro 2400Mhz
GPU: Powercolor PCS+OC R9 290

Scores:
Image Editing: 91111 Time: 94.17
Encoding: 83369 Time: 143.794
Open CL: 63941 KSamples/sec: 1562
Heavy Multi Tasking: 68851 Time: 142.045
System Score: 84029

Have no idea whether the score is good/average/great

Thoughts ?