Broadwell HD Graphics 5500 Delivering Similar Vulkan/OpenGL Performance

[phoronix]

Phoronix: Broadwell HD Graphics 5500 Delivering Similar Vulkan/OpenGL Performance

With having out a Core i7 "Broadwell" ThinkPad X1 Carbon laptop for the MoCA 2.0 network tests, I decided to run some end-of-year graphics tests on this Core i7 5600U system with OpenGL and Vulkan...

http://www.phoronix.com/scan.php?page=news_item&px=HD-Graphics-5500-EOY2016

[bisby]

The bars feel very misleading - using a different scale for openGL vs vulkan. After seeing 800x600 using the same scale and Vulkan being in the lead, at a glance, I assumed the other resolution comparisons would be the same (it would be nice if the openbenchmarking graph had a parameter to set the scaling) .

Would be even nicer if it could include all the comparisons with a single variable change into a single graph so the things comparing were right next to each other.

[dungeon]

Maybe people should look at numbers and not in lines

I readed this article but i am not interested to comment on "6 to 20 fps" articles, as that looks like out of this world Really playing games on iGPUs and APUs sometimes require tweaking some game settings and that is for what settings are for in PC games.

And for that Iris Pro NUC Intel even has page with recommended settings for popular games... altough for Windows But that is a point, as PC is not Playstation... further any PC is not faster than Playstation 4 PRO... hm, maybe only Titan X

https://gameplay.intel.com/HardwareG...03.50%20GHz%29

So as PC is about following recommendations, just follow recommendations there and Iris Skull would be fine... i guess

[bisby]

Maybe people should look at numbers and not in lines

Probably. my point was on ease of reading. it would have been nice to skim the article and see the relative difference. I dont even have a broadwell CPU, so I only care about the broad picture of OpenGL vs Vulkan.

further any PC is not faster than Playstation 4 PRO

Wait really? The PS4 Pro is using essentially a Radeon RX470. If we want to give PS4 "programming for a specific GPU" performance boost (which btw, isnt that great of a boost), we can assume it will perform roughly at the same level of an RX480 (on windows, since AMD and opengl dont play well). You definitely don't need a Titan X to outdo a PS4 Pro.

[dungeon]

Wait really?

Really. Titan X from previous gen i mean or even current, but with special point. GTX 1070 and bigger should be roughly clear faster on paper, but that is again not any PC - only some of PCs

Particualry not in this iGPU context plus need to follow recommendations on PC for particualar setup (for something like here which is not Titan X) point Since on Playstation PRO there is not much recommendations to follow needed - it is just buy and play

I think my point was clear, that with Titan X you don't mind about settings and everybody else probably should disable something here and there, particulary when someone wanna using some of those iGPUs for gaming That as i am pretty sure near nobody play games with 6 to 20 fps

[dungeon]

It is pretty simple on consoles and for PC, for majorty of titles targets are smooth and again smooth 30 fps or 60 fps usually... And of course on any iGPU or dGPU by definition majority of games can be played.

Now only question here for user (instead of user, Intel for some games for Windows already made some suggestions) on particular iGPU hardware is - on what settings we can (or not) match target framerate - so that we can claim something as "real playable"? But not, let say - what fps iGPU can get, when even Titan X can't do 60 fps

[torsionbar28]

Maybe people should look at numbers and not in lines

Defeats the purpose of a graph, no? A spreadsheet could convey the same numeric information.

[dungeon]

Defeats the purpose of a graph, no? A spreadsheet could convey the same numeric information.

When results are merged in one, graph scale well... it is just that two separate ones stand near each other, thus just unintentionaly making optical illusion, as always proved by looking at numbers