Is there more than CPU Benchmark Scores?

Paul22000

Distinguished
Mar 11, 2007
162
0
18,680
I currently have an AMD Phenom II X4 955 that I purchased in August of 2009. Yes, SIX years ago. It still works perfectly fine for all games, including Elite Dangerous and The Witcher 3. (I'm using a GTX 970.) The only problem is when I'm gaming with friends and if I share my screen on Skype, then the FPS drops in Elite Dangerous from a steady 60 to a choppy as heck 30.

Not just because of Skype, but I suppose after 6 years I can indulge in an upgrade. I'm considering upgrading to the i5-6600K as recommended for "high-end" in the latest Best CPUs article ( http://www.tomshardware.com/reviews/best-cpus,3986.html ).

However, this CPU only gets a score of 7740 compared to 4013 for my Phenom X4 955. ( http://www.cpubenchmark.net/cpu_list.php )

A) Does this mean that performance will ONLY be roughly twice as good? Kind of sad for 6 years of technological advancement to the latest and greatest CPU... Especially since the now "old" i7s are over 11,000. What is the deal?

B) Or is it possible that performance will actually be a LOT better because of the improvements in architecture, and faster RAM, etc? Basically I'm wondering if it's just a straight up comparison between 4013 and 7740, in which case I might wait for the 10,000+ scored CPUs to drop in price more, or if there's more behind the scenes that will actually make the upgrade worthwhile?
 

sammy sung

Distinguished
IPC advancement to roughly double that of a processor released six years ago sounds about right. In the initial case we're comparing quad core to quad core, making the core to core comparison even in terms of that bench. Whereas comparing either to an 8 thread i7 makes it a touch unfair.

The 6600K would be a beastly upgrade
 
Synthetic benchmarks need to be taken with a grain of salt. It's just one way to measure cpu performance, there are a ton of benchmark tests out there including other synthetics and a lot of real world tests. Real world benchmarks are more realistic since they use an actual program someone may use to show just what the difference is.

Some things to keep in mind for any benchmark test whether synthetic or real world, you have to compare the results from the same test. You can't compare scores from older passmark tests to newer ones since just updating the test code affects results. Much in the same way you'll get no meaningful comparison between two sports cars racing them on different racetracks. Passmark tries to assign points as a scale rating for comparisons but no explanation what the points are or what they mean. They're arbitrary. Cpu A is 1000 better than cpu B - 1000 what? 1000 minutes faster? 1000 oranges more? It greatly helps to have a scale that relates to something meaningful. Those scores don't relate to application processing times or fps performance in games but real world benchmarks will do just that.

An example would be the fx 8350 8 core which scores a passmark of 8970 points vs the i5 4690k which 'only' scores 7729 points. Looking at a cpu benchmark for adobe premier pro test encoding the same video clip, the 8350 took 139 seconds while the the 4690k took 135 seconds. According to passmark the 8350 is 16% faster in fictional point ratings and yet in a real program the i5 was about 3% faster.
http://www.tomshardware.com/charts/cpu-charts-2015/-31-Adobe-Premiere-Pro-CC,3722.html

Of course that's just one benchmark in one application, when it comes to gaming then game benchmarks need to be compared. One cpu may give the same fps on 2 different games while another gives 2 totally different fps values depending on the game. It just goes to show how far off base synthetic tests can be and what little meaning they have. In the above case if someone were looking for a video editing cpu using premiere pro passmark results would have grossly misled them into thinking the 8350 was much faster.

I think the i5 6600k would be a significant upgrade but for streaming you may want an i7 to take advantage of hyperthreading. Hopefully more people can chime in who have more experience streaming, usually streaming involves twitch, obs or similar but I'm even less familiar with skype and what I would think would be considered streaming via skype (uploading game play).

The internet connection will play a role in it for upload speeds, if the video is being processed/converted on the fly to be uploaded through skype then there's a chance the hard drive is also taking a performance hit. It's already trying to keep the game fed with data. Someone better versed in the details of streaming can likely give more detailed advice to get your fps up to optimal performance.
 

Shalmaneser

Distinguished
Sep 3, 2013
32
1
18,535

Beastly....as in great? Or big, hot, smelly, and overpriced?
 
If you'd like help regarding cpu's it would be better to start a new thread of your own Shalmaneser. Though if I had to guess at sammy sung's remark, beastly as in great. Intel cpu's especially the i5's and i7's are some of the most power efficient and strongest performing cpu's across the board. They consume far less power than amd 8 core cpu's which continue to fall further behind in terms of performance.

The skylake i7's like the 6700k are a bit overpriced due to lack of availability at the moment which means demand is greater than supply which drives price. Aside from that exception, both amd and intel cpu's are fairly priced according to performance they offer. If someone doesn't have a need a for a higher performing cpu it doesn't necessarily mean it's overpriced. It just means it's more than they need. For those who do it's not overpriced.