Consistently low scores on XTU with 4690K

[opticalgenesis]

Hi,
I've been everywhere trying to find a solution for this, but I've come up completely dry 🙁

I recently built my first PC, and chose a 4690K for OC ability. I only recently read about the Intel XTU so I benched my CPU at stock and got ~830 marks. So I OC'd to 4.2, and got 815. I've even tried 4.8GHz, and still no higher than 830 🙁

Do I have a bad chip? Or am I doing something wrong?

Please help!

Josh.

 

[GhislainG]

What were your expectations and what are other values (frequency, maximum temperature, thermal throttling, etc.)

I get 819 on a 4590 and an H97 chipset motherboard; I know that yours shouldn't be that different.

 

[opticalgenesis]

What were your expectations and what are other values (frequency, maximum temperature, thermal throttling, etc.)

I get 819 on a 4590 and an H97 chipset motherboard; I know that yours shouldn't be that different.

I accidentally marked your response as a solution...I apologise...

ANYWAY, at 4.4GHz (what I'm running now), obviously frequency is 4400MHz, max temps reported by BIOS is ~50 celsius, 0% thermal throttling and my cache is left as it is (3900MHz).

To answer your question on expectations, I've seen people, with setups not too dissimilar to mine, hit 1000+, and I would simply like to be able to at least breach 900 🙁

 

[synphul]

Hwbot scores are a bit weird and they're not directly related to performance in terms of speed like other typical benchmarks. They have their own algorithm that factors not only gigahertz but vcore as well. I haven't figured it out exactly but say you're running 4.4ghz at 1.25v vcore and someone else is running 4.4ghz at 1.18v vcore, they will earn a higher score than you will. It doesn't mean their cpu is any faster than yours. Another odd thing with hwbot scores obtained from xtu is how reliant it is on software tweaks. When I first built my system and reinstalled win7 x64 ultimate I was having fun messing with my cpu and overclock settings, grabbed xtu and was curious to see how it rated. I scored like 415.

When I saw everyone else's score I was really bummed out. Then found where it suggested I upgrade to sp1 - I'd installed windows but hadn't installed sp1 yet. Re-ran the test and immediately hit just over 800 yet I highly doubt my cpu magically doubled in performance purely from installing sp1. Just goes to show how some rating tests are heavily software dependent and don't solely measure hardware performance.

As with any benchmarking, take it with a grain of salt. Meaningless numbers from synthetic tests don't count for much, what counts are real world tests like how long it takes to zip a set batch of files or how long it takes to render the exact same image in photoshop etc.

 

[GhislainG]

I unselected the best answer. Have you compared your score with similar systems (same motherboard and similar memory)? To be honest mine seems to match other 4590's, but I don't rely on such a test. I can't overclock, therefore I don't know if it would increase the score or not. What cooler do you have to only reach 50°C at 4.4 GHz? The best score for an i5 4690K is 1397 at 5616 MHz using LN to cool it, a Maximus VII Impact motherboard and a 1200W PSU.

 

[RuinedXJake]

I just got 927 at 4.2 GHz clock with 1.105V vcore and 3.9GHz cache and auto voltage.
Something is underperforming in your build, I don't know what though.