I've built rigs for the 25 years since I was 12. They're usually very high end rigs, where phenomenal performance [i]and[/i] reliability are vital.
My tool-set changes and expands with the landscape. On Windows, some of my favorite utilities remain the following, offered for informative purposes:
+ SpeedFan
+ CPU-Z
+ GPU-Z
+ memtest
+ RealTemp
+ [url=http://www.ocbase.com]OCCT[/url]
These are robust and [i]free[/i]. They focus on hardware monitoring, testing, benchmarking, and burn-in. I prefer pegging-out my CPU, RAM, GPU, and PSU for at least 24h before any overclocking or tweaking. Once I'm happy with a configuration, I then burn-in for as many days as patience allows. If something's going to die, I want to know while I can easily RMA it.
Modern GPU manufacturers also distribute good utilities with their hardware.
[b]----[/b]
Utilities like Catzilla and 3DMARK (an established alternative I'll discuss alongside Catzilla) provide benchmarks geared toward replicating possible realistic gaming usage rather than stressing and killing your system.
Catzilla performs a similar set of complex graphical scene tests as others in this field and provides you an overall score with a breakdown of test elements at the end and an option to submit to an online database for comparison to the wider testing population.
It also seems to use the same score calculations. The numbers come in similar to tests on other suites and at the same "better than 90%" placement that other suites give this 2.5yr old system.
Core functionality, therefore, seems on-par with competitors.
It is everything wrapped around that core which erodes my interest.
[b]----[/b]
Launching Catzilla hung my system and reduced the GUI to low-memory mode for the [i]two minute[/i] load time. Competing products launch instantly.
Once launched, the first thing I noticed was that they used resolution terminology that put me off, because it comes across as inexperienced. They provide only the vertical number, followed by a "p".
It may seem pedantic, but when referencing display resolutions in the world of PCs, you provide both horizontal and vertical measurement (1920x1200). We do this, because there are many aspect ratios (5:4, 4:3, 16:9, 16:10, 21:9). Providing a single number omits vital info that can impact calculations like estimated performance or configuration.
You can't assume 1080 vertical means anything, because horizontal could be 1920 at 16:9, but 2560 at 21:9. You can't assume 2560 horizontal means anything, because at 16:9 vertical would be 1440, but 1600 at 16:10.
This terminology is acceptable for TVs, because they're standardized at 16:9. While prevalence of 16:9 TVs made it common with computers (manufacturers would rather produce one ratio than reconfigure plants for multiple), you can't count in this being the case.
We also don't add [b][i]p[/i][/b] to the resolution. It stands for [i]progressive scan[/i] (as opposed to [i]interlaced[/i]), which every monitor inherently is, so there's no point going out of our way to redundantly state this.
This is accepted in TVs, because some TVs only do interlaced at certain resolutions and some content is only progressive or interlaced and some hardware (like last gen consoles) only deliver an interlaced signal, in some cases.
I can accept someone coming from the console world, with little PC experience, or simply caught up in the habit of using TV-specific terminology may discuss resolutions in this way, but it leaves me with a negative impression from a [i]professional software developer[/i] putting out PC software to test PC displays for PC rigs.
3DMARK and other suites give full resolution detail when referenced and uses accurate terminology.
[b]----[/b]
The second negative impression came with trying to customize a test for my system. The default tests run at 576 (oops, "576p"!), 720, 1080, and 1440. All 16:9 ratios. I have a 16:10 display and I want a benchmark that is representative of what I'm going to have when I'm gaming at 16:10 at a native resolution.
Unfortunately, you can't. Even "custom" configuration does not allow you to choose anything but those four resolutions. All you can customize are options like multi-threading. A developer confirms in their forums that there is no way to customize resolution and this will never change.
Presumably, this is to standardize certain parameters for comparison of your benchmark against others in the database. Fine. I still want the option to perform my own benchmark under my own conditions and parameters so that I can compare my system against itself after performance tuning. These four resolutions make for a constrained testing environment.
3DMARK and other suites allow full configuration of these options and even uploading them to their database for comparisons.
[b]----[/b]
I was disappointed that the touted Hardware recommendation feature/tab never worked. It had zero results and kept saying that I needed to run a full test. This was supposed to be a feature which links you to advised hardware upgrades.
[b]----[/b]
Another problem you will encounter is navigating their site. A dark and "edgy" interface which looks more at home as the official Call of Duty site. It isn't enjoyable to use. In fact, it can be confusing.
One button pulls up a set of information, but it seems broken, until you realize the data has been displayed a couple screens downward, without any evident notification.
Also, one expects the "compare to other users" button to match your results against the database, until much frustration ultimately reveals it only compares you against one user that you must provide a name for. Comparison against the whole population requires going back to the application and selecting "show all" under the Top 10 users list, which launches your browser and takes you to the database.
[b]----[/b]
Worse, you have to register an account and link it to your application. An overly complex process involving multiple long codes, contradictory sets of instructions, and numerous failed attempts. I ultimately had to rely on a user in the forums who found another guy's steps listed somewhere. Even these did not help; but they guided me in the right direction. This should be a simple matter of "application gives me a code, I go to website and enter code on my profile". Some parts of the process even take you to dead URLs with error messages.
My final complaint is that when you exit the program, Steam thinks it is still running. I am unable to identify this application running in the process table, but Steam will not close, because of it. You can not launch Catzilla again, either, until you reboot your computer.
[b]----[/b]
Ultimately, Catzilla is serviceable.
It has received some press in the last year, but it's usually around how quirky the superfluous test video is. Dismissing that; focusing on what is actually provided, it does what other packages do, but in a less polished way with fewer customization options.
I still prefer 3DMARK, which I mention in this review because of its popularity, my familiarity, and its availability on Steam. While $10 USD more, it's more robust, less glitchy, has a pleasing interface, allows more customization and control, and has an established history and reputation. It is also frequently updated and offers a wider variety of specialized tests for everything from a tablet, notebook, and laptop to an extreme gaming rig.
Even better, wait until 3DMARK is on sale. I've seen it 90% off ($2.50) in the last week of 2013. While waiting for a sale, use the free version (or free version of another suite).
But keep an eye on Catzilla. I love competition and hope they continue to refine this. I hope it evolves into a robust utility giving competitors a run for their money. I just don't see the current iteration being part of my tool-set at the current price.
