My Benchmark Is Terrible...

[phil.r.greco]

Hi all, given the PC was built in 2012, it can still handle mainly everything i throw at it, some with performance issues obviously. I ran the benchmark and below are the scores. I plan on building a new rig, but curious as to why it scored the computer as basically a boat anchor for the desktop/workstation scores. The RAM is maxed for this motherboard, although one problem i did notice is i purchased 1066 instead of 1600 for memory speed. I plan on passing this PC to the wife once i build another with the aid of the benchmark website, but any thoughts on how this one can possibly improve on that score? or, is the benchmark website full of it LOL

UserBenchmarks: Game 23%, Desk 33%, Work 25%
CPU: AMD Phenom II X6 1045T - 40.4%
GPU: Nvidia GTX 650 Ti Boost - 24.3%
HDD: Seagate Barracuda 7200.14 500GB - 43.8%
HDD: WD Blue 500GB (2008) - 53.8%
USB: Seagate Backup+ Hub BK 6TB - 19.4%
RAM: Unknown 2x4GB - 31.3%
MBD: Asus M4N68T-M-V2

 

[bigpinkdragon286]

Most benchmarks are full of it. That one in particular at least gives you amusing names for their irritating scores. It's basically a site that tends to give good scores to ridiculously overpriced builds, and casts shame on decent, but otherwise aging equipment.

Don't listen to the nonsense about bottlenecks, unless you like running in circles. All computers have bottlenecks always, and the bottleneck changes on identical equipment based on context. My favorite quote I've heard someone give for the GPU bottleneck question is, "Yes, no, maybe, sometimes, and it depends." That's really the best answer for bottlenecks, despite sites like the one you mention touting to know best how big your bottleneck is, which can't actually be known until you have the system put into context, which requires specific settings and specific software.

What needs to be improved in your old system? If it does everything you want, ignore the benchmarking site. If a particular aspect needs improvement, upgrade that, and continue to ignore that silly site. There's no way you're going to get good scores on that site by tinkering with parts that old.

 

[phil.r.greco]

Thanks, i figured it was full of it due to the fact, the machine does quite well, even for newer games that demand newer gear. I obviously just cannot max out the settings. I plan to build another soon, but with 3 kids and one in college, the funds aren't always there. Thanks for your feedback!

 

[USAFRet]

or, is the benchmark website full of it LOL

Yes.

Does the PC do what you need it to do? If so, don't worry.

 

[phil.r.greco]

It does, but very sluggish under load for the one game i find time to play sadly... but oh well... i'll have to wait until i can build something faster.

 

[bigpinkdragon286]

The 1045T was never a fast CPU, even at release. It made up for lack of sheer speed in some cases with it's extra cores. The cost of dropping in a 1090T or 1100T is more than I would consider worthwhile for such an old system, and those CPUs are still going to be less than stellar, speed wise. You would see more return on money from even a bottom rung Ryzen CPU/MB/RAM swap. You'll probably find out that the 650 Ti is the major issue at that point.

The 650 Ti was never considered more than mid-tier, and that was about 5 card cycles ago.

Honestly, it's only my opinion, but I think the system looks reasonably well balanced as is. Before you rush to upgrade to play just one game, make sure your upgrade is going to benefit you. Some games run like potatoes even on high end equipment.

If I were trying to add a little more oomph to it, I would scour my local Craig's list market for a decent, used GPU, and leave it at that, maybe spend some time on eBay, but honestly, eBay looks like a potential disaster area for GPU deals right now. Barring that, I would consider a new GTX 1050 or RX 560. I doubt I would shoot any higher than that myself unless I was planning to reuse the GPU in a better system later.

Edit: I suspect even the R3 2200G with it's integrated Vega 8 GPU will return similar performance to your GTX 650 Ti.

 

[Aeacus]

Reason why your UserBenchmark score is so low is because compared to the best hardware currently available, your PC is weak, hence the low score.

Also, i don't think UserBenchmark is "full of it" by giving the results based on face value. I'll take my Skylake build as an example here, full specs with pics in my sig.

My 1st run at UserBenchmark gave me these results:

UserBenchmarks: Game 72%, Desk 80%, Work 49%
CPU: Intel Core i5-6600K - 93.2%
GPU: Nvidia GTX 1060-3GB - 72.4%
SSD: Kingston HyperX 3K 240GB - 78.5%
HDD: WD Blue 500GB (2010) - 37.6%
HDD: WD Blue 1TB (2012) - 62%
USB: Kingston DataTraveler 2.0 8GB - 4.3%
RAM: Kingston HyperX DDR4 3000 C15 2x4GB - 101.6%
MBD: MSI Z170A GAMING M5 (MS-7977)

Not bad, eh?
Over the time, i've upgraded my Skylake build and also ran additional runs.

2nd run was done once i upgraded my RAM to 16GB and replaced aging 500GB HDD with 1TB HDD:

UserBenchmarks: Game 72%, Desk 82%, Work 55%
CPU: Intel Core i5-6600K - 90.7%
GPU: Nvidia GTX 1060-3GB - 71.7%
SSD: Kingston HyperX 3K 240GB - 74.6%
HDD: WD Blue 1TB (2012) - 97.8%
HDD: WD Blue 1TB (2012) - 63.4%
RAM: Kingston HyperX DDR4 3000 C15 4x4GB - 105.2%
MBD: MSI Z170A GAMING M5 (MS-7977)

With RAM upgrade, i gained 2% on Desk stat and 5% on Workstation stat. Game stat didn't increase since the formula used to calculate overall stats doesn't consider RAM and HDD capacity as important contributors.

3rd run was done when i replaced my OS drive: 2.5" SATA SSD with M.2 NVMe SSD:

UserBenchmarks: Game 79%, Desk 129%, Work 76%
CPU: Intel Core i5-6600K - 92.7%
GPU: Nvidia GTX 1060-3GB - 71.3%
SSD: Samsung 960 Evo NVMe PCIe M.2 500GB - 291.3%
HDD: WD Blue 1TB (2012) - 89.5%
HDD: WD Blue 1TB (2012) - 67.1%
RAM: Kingston HyperX DDR4 3000 C15 4x4GB - 106%
MBD: MSI Z170A GAMING M5 (MS-7977)

Due to the incredibly fast read/write speeds of M.2 NVMe SSD, i gained 7% in Game stat, 21% in Workstation stat and whopping 47% in Desk stat.

4th run was made after i replaced one of my aging 1TB HDDs with 1TB SSD:

UserBenchmarks: Game 80%, Desk 131%, Work 77%
CPU: Intel Core i5-6600K - 93.2%
GPU: Nvidia GTX 1060-3GB - 71.7%
SSD: Samsung 960 Evo NVMe PCIe M.2 500GB - 297.4%
SSD: Crucial MX500 1TB - 78.8%
HDD: WD Blue 1TB (2012) - 92.4%
RAM: Kingston HyperX DDR4 3000 C15 4x4GB - 106.2%
MBD: MSI Z170A GAMING M5 (MS-7977)

Changing storage drive to a faster one gave me additional 1% in Game, 1% in Workstation and 2% in Desk stat.

When looking the bench scores throughout my upgrade progress, i can't say that UserBenchmark is "full of it". Also, if i were to upgrade my GPU to a e.g GTX 1070 Ti, i'd gain few % in Desk and Workstation stat and about 35% in Game stat. Though, for time being, i don't plan to upgrade my GPU and i'm happy with my PC.
Here's further reading on how the UserBenchmark score is calculated,
link: http://www.userbenchmark.com/Faq/What-are-the-UBM-performance-classifications/93

If my plain text scores aren't sufficient proof, here are links to each of my runs as well:
1st run: http://www.userbenchmark.com/UserRun/3416648
2nd run: http://www.userbenchmark.com/UserRun/3760877
3rd run: http://www.userbenchmark.com/UserRun/7828840
4th run: http://www.userbenchmark.com/UserRun/8749818

 

[bigpinkdragon286]

I find it amusing how the benchmark scores are never consistent, even for the same piece of equipment.

I guess I'm the only one that finds it problematic when benchmarking gives results based on a moving reference.

 

[Aeacus]

There are different variances in each bench run and it's impossible to get identical results each and every time.

If you don't trust UserBenchmark, take for example Unigine Superposition and run the 1080p High preset couple of times. Look if you can get the identical results in each and every run on your machine. I'm quite sure you won't. I didn't.

I, personally, don't mind the slight variances in each of my component bench scores. I use UserBenchmark to detect component underperformance within my system.

 

[cryoburner]

It's basically a site that tends to give good scores to ridiculously overpriced builds, and casts shame on decent, but otherwise aging equipment.

Those numbers actually seem reasonable to me. They're simply comparing the hardware against modern, higher-end equipment. For example, CPU performance is currently compared against a 7700K and GPU performance is compared against a GTX 1070. Provided their performance is not getting limited by other components, a GTX 1070 can be several times as fast as a 650 Ti, and a 7700K is more than twice as fast as an X6 1045T. Of course, not everyone needs that kind of performance, and a score less than "100%" doesn't mean that the computer won't be able to serve those purposes, just that it's below the cutoff point for what they consider "high-end".

UserBench provides a lot of useful information, allowing one to easily tell roughly how components compare against different hardware, and to verify that their components are not performing significantly below where they should be. Personally, I think the percentage numbers for gaming, desktop, and workstation use are a bit silly, as the resulting number is an abstraction generalizing what the system might potentially be getting used for. You have to dive down into the details to find the more useful information. For example, the graphs that compare each component's performance against benchmarks from other users with the same piece of hardware, or the charts comparing your hardware against a potential upgrade you might be considering.

I found your actual benchmark run, by looking in the recent user-submitted results for that CPU...

http://www.userbenchmark.com/UserRun/9937606

From looking at the graphs, CPU and GPU results appear to be right in line for those particular components, and the same goes for your 500GB WD drive. The 500GB Seagate system drive appears to be performing a little slower than it should be, though that might have potentially just been due to another program, or the OS, accessing the drive at the time you ran your test.

The 6GB external drive is clearly having its performance limited by being connected to a USB 2.0 port. If you have a USB 3.0 port on that system (likely colored blue) you would get a lot better performance from that drive by plugging it in there. If not, you could always add some USB 3.0 ports, if you're motherboard has an empty PCIe slot, and if you consider the performance of that external drive to be important. There are a number of USB 3.0 cards priced around the $15-$20 range...

https://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=100160874%208000%20600022031%20600437850%20600022010%204814&IsNodeId=1&bop=And&Order=PRICE&PageSize=36

 

[USAFRet]

The thing I don't like about userbenchmark is the conflicting data it gives for the same part at the same time.

Very good and 16th Percentile Performing below expectations!

My CPU, just now:

People glom on to that "way below" text and freak out. Ignoring the "Very good".


And my GPU:

CPU at 76.5% (way below)
GPU at 73.6% (way above)

So which is it? Confusing.

 

[cryoburner]

What they're saying is that CPU performance is "very good" relative to all other systems, but "below expectations" compared to other 4790Ks. That graph is showing benchmarks from other 4790Ks, and you can clearly see that "this bench" is near the bottom of the curve. So while 76.5% is considered very good, it's below the average score for a 4790K of 84%. Maybe it's running too hot to boost properly, or turbo boost is disabled. Looking at the results, "base clock 4GHz, turbo 4GHz" seems to indicate that might be the case, since a 4790K should boost up to 4.4GHz on one core, or 4.2GHz on all cores. This probably wouldn't be very noticeable, since it works out to only a 5-10% reduction in clock speed compared to stock, but it's something that might be worth looking into, which is why they point it out.

For your RX 580, the opposite is true, and your card is performing above the average results for that card. It scored 73.6%, compared to 69% as the average. If you click on the little question mark next to the graph, it will take you to a page explaining that the number relates to other tests of that particular piece of hardware, while the one above the number to the right points out that the score is "relative to the best alternatives", which they arbitrarily decided to be a 7700K and a GTX 1070.

Perhaps the descriptions of what the numbers are referring to could be be presented in a more obvious way, but there's already a lot of information being displayed there. UserBench might not be perfect, but I'd say it's currently one of the best sites for what it sets out to do.

 

[USAFRet]

Perhaps the descriptions of what the numbers are referring to could be be presented in a more obvious way, but ...

I know that, and you know that.
But almost daily, we see people here moaning about "Way below...", when in reality, it is running just fine.

Consider this:

Let's assume 100% is absolute best, out of the factory performance.
All the parts tested range between 90 and 95%.
All still very, very good, right?

Your CPU/GPU/SSD at 90.8% will be displayed as "Performing way below expectations 19th percentile"
Bottom fifth of all the parts tested, but all within a very narrow range of "Very good".

Confusing.