I really enjoy and trust HH GPU reviews but I would prefer it if the fastest none reference was included in the review tables as on this forum that's the cards we buy. Showing the reference as in the 980ti when you compare it to a none reference the difference is huge the none reference is more like a 990ti they overclock better there cooler and much faster in every game. I know having all none reference would be to much but when a new gen GPU comes out seeing the best of the old gen GPU in low/med/high would give us a real world comparison. I use Guru3d to choose my next GPU but only using reference is making that very hard.
Reference is the most objective point HH can probably get. Since nVidia changed GPU boost every non-reference is like a dice roll as per clocks/performance. Too much work to get consistent results.
I agree including none reference is difficult but when reference is so far from none reference it is pointless including it as it does not show the true GPU performance. That's why I said the best of none reference that way we see the best of the old gen against the new gen
There are way too many versions on the non-reference cards to be tested too. EVGA alone has like 8 different versions of the 980ti.
For the broadest appeal, would think the best selling card of past generations stay. That way one can see what the installed performance base is like and what kind of performance increase there is to that base. With keeping the reference cards in the picture, one can see what was initially released and what the public actually acquired from that generation. EDIT: the way the reviews currently are is great. There is a good mix and one is able to extrapolate to some degree where a particular version would fit. It's all for a generalized picture anyway.
My point is it would take Hilbert 2-3x as long to test them all and that's if he even gets each card. I'm sure the GPU makers don't send him 1 of each version of their cards so how is he going to test them?
Benchmarks for the 980ti have already been done. But only the reference is included in the new 10 series GPU reviews. I figure when the same game or benchmark is there why not include the best of the old GEN a simple copy and paste of the results into the chart. An example would be latest 1070 strix review show 3dmark score of just over 15656 for a 980ti but the card I have in Guru3D 3dmark review score over 17235 stock 18858 oc that's a huge difference nearly all none ref 980ti in that bench score higher than the 1070 the strix scores 16822
All information taken from Guru3D hardware reviews 3Dmark 2013 Fire Strike.....STOCK.....OC Zotac GTX 1080 AMP Ex.........20312 REF Geforce GTX 1080.......19370 GBT GTX 980ti WATERFORCE..17916.....19178 MSI GTX 98Ti Lightning..........17853.....19011 MSI GTX 980Ti Gaming 6G.....17325.....18858 Asus GTX GTX 1070 STRIX.....16822 Asus GTX 980ti Poseidon........16649.....18816 REF Geforce GTX1070 .......16229 REF Geforce GTX 980Ti......15656 Radeon R9 Fury X.................14374.....15226 MSI GTX 980 Gaming............11831.....13280 MSI GTX 1060 6GT OC...........11729 ASUS GTX 980 Poseidon.........11872.....13722 REF Geforce GTX 1060.......11701 REF Geforce GTX 980.........11168 Palit GTX 970 Jetstream.........10044.....10901 REF Geforce GTX 970...........9568
I understand your suggestion, but reference is the baseline, the rest is offset. If you want to look up the results for a tweaked card look at the specific review on that ? If I'd test the fastest 980 Ti and compare it to the fastest 1070 how would that be any different comparing a reference 980 Ti to a 1070 reference in relative percentage wise ? Also by default you assume that everybody purchases the fastest card. That just makes no sense whatsoever. By just testing the fastest clocked SKUs out there we'd be part of the smaller group. Most people will buy close to reference based on price. Not saying I don't understand your point of view here, but you are asking to compare apples and oranges just because you feel treated a bit unfair as your faster 980TI is not being compared. Also the data-sets would get confusing, very confusing with all the brands and SKUs. Charts need to be crispy clean to be able to grasp and understand them, that's what reference results offer. Anyone with a little common sense understands that when he bought a more expensive fast SKU and compares it towards new reference baseline products there will be a difference. Just the fact that it feels a little unfair as you are on a highly tweaked 980 Ti doesn't justify the request to test the fastest SKUs only. BTW I am moving towards to adding an extra benchmark that compares the GPU 3DMark score in-between all the SKUs so that part of what you ask here, is adressed.
Sorry to have caused you extra work the extra benchmark addresses everything for me I spend a lot of time looking through old reviews to get a real idea of performance when looking at upgrading. The 980ti ref is so different to the none reference 980ti I could not decide if upgrading was an option without looking through all the old reviews. Your reviews are the ones I trust and base my furture purchases on thanks HH