LTT 3DMark Thread

[done12many2]

1 hour ago, Kryptyx said:

So I'm thinking about putting emphases on the Graphics score over the combined score.

 

How would you go about this?  You could make a tab with graphic scores only for each test.

 

I personally think that overall score has value as designed by 3DMark.  Different setups will favor higher graphics score, while others with favor higher CPU scores.  As @DXMember taught me, it's about the balance with regards to combined score.

 

 

[Vellinious]

That's true, if you care about how much money someone spent on a processor....I mean, I'd leave the overall scores on the sheet, but I'd have it sorted by Graphics score.  After all, we're measuring video card performance here....the 3D Mark Physics test isn't a very good CPU benchmark.  

I think it'd level the field a bit, actually...no more top scores from guys with $1000 processors riding their killer physics scores to the top of the list, while their GPU scores are less than stellar, and usually bordering on average.

[done12many2]

38 minutes ago, Vellinious said:

That's true, if you care about how much money someone spent on a processor....I mean, I'd leave the overall scores on the sheet, but I'd have it sorted by Graphics score.  After all, we're measuring video card performance here....the 3D Mark Physics test isn't a very good CPU benchmark.  

I think it'd level the field a bit, actually...no more top scores from guys with $1000 processors riding their killer physics scores to the top of the list, while their GPU scores are less than stellar, and usually bordering on average.

 

Why am I looking over my shoulder right now?   haha

 

I personally focus on graphics scores more than physics, but I also believe that you should tweak your rig to the benchmark.  Not tweak the benchmark to your rig.  By picking one aspect of measurement, I feel we're picking the part of it that works better for us.  

 

What we currently do maintains the spirit of the test.  If we want to modify that, it should be an additional informational column and not the focus.

 

Those guys riding on those killer physics scores are exposed in other benchmarks and I think we can all see that.

[arnavvr]

That's the only reason I want a E5 1680 V2. But they are still $1000..

[Vellinious]

3 hours ago, done12many2 said:

 

Why am I looking over my shoulder right now?   haha

 

I personally focus on graphics scores more than physics, but I also believe that you should tweak your rig to the benchmark.  Not tweak the benchmark to your rig.  By picking one aspect of measurement, I feel we're picking the part of it that works better for us.  

 

What we currently do maintains the spirit of the test.  If we want to modify that, it should be an additional informational column and not the focus.

 

Those guys riding on those killer physics scores are exposed in other benchmarks and I think we can all see that.

Firestrike has always been a graphics benchmark....at least, that's how I've always used it.  Physics is an afterthought for me, but....gotta have it, if you want a decent score.  Just seems a tad nonsensical to me.  It's like running your car around the race track to test your taillights....

[arnavvr]

1 minute ago, Vellinious said:

Firestrike has always been a graphics benchmark....at least, that's how I've always used it.  Physics is an afterthought for me, but....gotta have it, if you want a decent score.  Just seems a tad nonsensical to me.  It's like running your car around the race track to test your taillights....

 

I always viewed Firestrike as a whole system benchmark. Have you looked into a E5-1660 V3?

[Vellinious]

Just bought a 6950X a few minutes ago.  With the graphics scores I hit, I should easy hit the Hall of Fame in a few benches.

Can't beat em, join em then beat em.

[done12many2]

27 minutes ago, Vellinious said:

Firestrike has always been a graphics benchmark....at least, that's how I've always used it.  Physics is an afterthought for me, but....gotta have it, if you want a decent score.  Just seems a tad nonsensical to me.  It's like running your car around the race track to test your taillights....

 

There's really no sense in debating what we consider FireStrike to be.  It was designed with graphics, physics and combined scores for a reason.  Overall performance.  

 

I have always considered the 3DMark series a tool to measure overall system performance.  Graphics scores are commonly used in reviews for card performance and physics scores are commonly used in CPU reviews.  It's a well-known standard of comparison.

 

Our individual opinion of the benchmark shouldn't influence the way we handle scores.

 

I personally don't like how Valley requires insane clocks speeds in order for me to max out 1080p performance, but that's how the benchmark runs.  With that said, I adjust to the benchmark and make my CPU do things it probably shouldn't.  

 

I like the the headlight analogy and can definitely see what your saying.  However the physics test isn't quite the same thing as headlights to racing. In racing, sometimes you need headlights and sometimes you don't.  In guaging overall system performance, you always need the CPU.  

 

I'd imagine this is why the designers repeatedily include the physics test generation after generation.  They obviously continue to lean towards overall system performance.

 

 

[done12many2]

7 minutes ago, Vellinious said:

Just bought a 6950X a few minutes ago.  With the graphics scores I hit, I should easy hit the Hall of Fame in a few benches.

Can't beat em, join em then beat em.

 

Okay.  Disregard everything I said today.  Let's do graphics scores only.  Haha.

 

Congrats bud.  You are about to dominate the physics scores.  

 

Even knowing that, I still believe we should leave the subs as is.

[PianoPlayer88Key]

On 7/14/2016 at 2:03 PM, Jumper118 said:

(re: TimeSpy)  not running it then also you need to turn off integrated graphics to get a correct score. 

 

Hmm... I was trying to run it on my desktop PC, and was getting no score. I get around 750-780 in FireStrike.  CPU is an Intel i7-4790K, GPU is Intel HD 4600.  I'm eventually planning to upgrade to a 1060 (or if they come down to MSRP $379 soon, a 2-fan 1070).

 

Running it on my laptop, with an i3-6100 & GTX 970M, yielded a little over 2100 in TS.  (FS usually does around 6000 or so.)

 

Hey does any good mult-generation, multi-platform overall system benchmark exist?  I'd like to see how much improvement my PCs are over my dad's Core2Duo 2008-vintage laptop, my bro's old AMD K6/7/whatever & GeForce FX 5200 system if it will still turn on (has been unplugged >10 years), his Pentium II-233 if he still has it, and if I ever laid my paws on one again, a 486, 286 (like my dad started with in 1989) or 8086-based PC or other pre-x86 systems.

[Jumper118]

6 hours ago, PianoPlayer88Key said:

Hey does any good mult-generation, multi-platform overall system benchmark exist?  I'd like to see how much improvement my PCs are over my dad's Core2Duo 2008-vintage laptop, my bro's old AMD K6/7/whatever & GeForce FX 5200 system if it will still turn on (has been unplugged >10 years), his Pentium II-233 if he still has it, and if I ever laid my paws on one again, a 486, 286 (like my dad started with in 1989) or 8086-based PC or other pre-x86 systems.

nope. 

[grayperview]

My R9Fury with fully unlocked shaders v Stock

 

http://www.3dmark.com/compare/fs/9470556/fs/8815933

 

1120Mhz Core 520Mhz Memory.

[DildorTheDecent]

6 minutes ago, super_skank said:

My R9Fury with fully unlocked shaders v Stock

 

http://www.3dmark.com/compare/fs/9470556/fs/8815933

 

1120Mhz Core 520Mhz Memory.

That's a nice 12% GPU score gain by not having to do much.

 

nice. 

[DXMember]

20 hours ago, Vellinious said:

That's true, if you care about how much money someone spent on a processor....I mean, I'd leave the overall scores on the sheet, but I'd have it sorted by Graphics score.  After all, we're measuring video card performance here....the 3D Mark Physics test isn't a very good CPU benchmark.  

I think it'd level the field a bit, actually...no more top scores from guys with $1000 processors riding their killer physics scores to the top of the list, while their GPU scores are less than stellar, and usually bordering on average.

guys talking about graphics scores being more important - you can sort the benches by graphics score if that's your thing... the sort button is there, you know...

[DXMember]

2 hours ago, super_skank said:

My R9Fury with fully unlocked shaders v Stock

 

http://www.3dmark.com/compare/fs/9470556/fs/8815933

 

1120Mhz Core 520Mhz Memory.

sick!

do you get any artefacts?

[grayperview]

1 hour ago, DXMember said:

sick!

do you get any artefacts?

None as of yet....only run benchmarks in a few games and 3d mark stresstest....and played the UAC level on doom (should stress shaders in theory)....i was up all night so I'm a bit too tired to game right now, but later i'll play some games.

[DXMember]

33 minutes ago, super_skank said:

None as of yet....only run benchmarks in a few games and 3d mark stresstest....and played the UAC level on doom (should stress shaders in theory)....i was up all night so I'm a bit too tired to game right now, but later i'll play some games.

damn... I can unlock 60 CUs on both, but it throws crazy artefacts once I unlock all 64

I think I can do 63 CUs on one card and 62 CUs on the other though

[grayperview]

4 hours ago, DXMember said:

damn... I can unlock 60 CUs on both, but it throws crazy artefacts once I unlock all 64

I think I can do 63 CUs on one card and 62 CUs on the other though

i got lucky i suppose

[Kryptyx]

7 hours ago, DXMember said:

guys talking about graphics scores being more important - you can sort the benches by graphics score if that's your thing... the sort button is there, you know...

The values are all null on the spy results. Personally I hate microsoft's online suite so I might opt to finish what I started before. I'm just thinking the default sort should be by graphics score since this is a graphics thread. I think those people who are able to push their GPUs the furthest shouldn't be behind people with bigger CPUs just because of the Physx score because in reality those people with the "lower" overall score (but have higher gpu score) will yield higher FPS in most circumstances which is what matters the most.

[done12many2]

17 minutes ago, Kryptyx said:

The values are all null on the spy results. Personally I hate microsoft's online suite so I might opt to finish what I started before. I'm just thinking the default sort should be by graphics score since this is a graphics thread. I think those people who are able to push their GPUs the furthest shouldn't be behind people with bigger CPUs just because of the Physx score because in reality those people with the "lower" overall score (but have higher gpu score) will yield higher FPS in most circumstances which is what matters the most.

 

I obviously disagree, but understand that it's your thread.

[Kryptyx]

1 hour ago, done12many2 said:

 

I obviously disagree, but understand that it's your thread.

Well I don't really think this is "my" thread as it is more for the community as a whole. I just happened to be the one who started it. I guess I was just trying to see what everyone else thought. 

But if you think about it, when comparing graphics cards and their performance the CPU gets in the way. You could have a higher overall 3dmark score but much less FPS in gaming applications.

[Vellinious]

I've always viewed the 3D Mark benchmarks as GPU benches.  I've never really cared about the overall scores....money wins that game.  Just my opinion.  /shrug

[DXMember]

5 hours ago, Kryptyx said:

The values are all null on the spy results

yeah, time spy needs a fix - I'll look into it later

the rest of the tests are fine though

 

3 hours ago, Kryptyx said:

But if you think about it, when comparing graphics cards and their performance the CPU gets in the way. You could have a higher overall 3dmark score but much less FPS in gaming applications.

While this is true to some extent, Futuremark will never allow a GT 710 with 6950X to beat a 750ti with 6500K because the total score is exponential average

if one of the scores is really imbalanced it will put the over score much lower to be more in line with the lowest score

it does however hold back crazy SLI and CF setups regardless of CPU used

[arnavvr]

@DXMember Why do you choose to use Office instead of Google Drive? 

[DXMember]

5 minutes ago, arnavvr said:

@DXMember Why do you choose to use Office instead of Google Drive? 

I honestly don't see a difference between the two for the end game - as in to display the results

MS is more convenient for me as I can use desktop Excel and publish it directly, it also offers more stuff to do and I'm more familiar with it