Re: TWC Total War Warhammer benchmark thread - last updated: 16.10.2016
Updated the results section and drivers section like described in the changelog (Original Post).
@Spartan many thanks for your participation. 2 question arised for me while adding your results. Is there a reason why your main rig is not yet running on the anniversary update? Could you be bothered to run a DX12 test with the 970 GTX rig?
Re: TWC Total War Warhammer benchmark thread - last updated: 16.10.2016
Originally Posted by alQamar
Updated the results section and drivers section like described in the changelog (Original Post).
@Spartan many thanks for your participation. 2 question arised for me while adding your results. Is there a reason why your main rig is not yet running on the anniversary update? Could you be bothered to run a DX12 test with the 970 GTX rig?
I rolled back from Anniversary Update because it nuked Steam. Basically couldn't launch any Steam games after the update.
I've since taken the Anniversary Update and it's working fine, so not sure what was going on there. I'll revisit DX12 on the GTX 970 rig when time permits. It has become my wife's gaming rig and I'd hate to kick her off Overwatch to run tests.
Re: TWC Total War Warhammer benchmark thread - last updated: 26.12.2016
Hi Druout thank you very much for this massive submission and tons of time you have invested here. How come that you have so many different systems at hand?
Re: TWC Total War Warhammer benchmark thread - last updated: 26.12.2016
Originally Posted by alQamar
Hi Druout thank you very much for this massive submission and tons of time you have invested here. How come that you have so many different systems at hand?
Well, I really enjoy PC hardware, so much so that I haven't bought a new rifle or pistol in about five years now so I can spend that money on parts. I didn't start building my own computers until a few years ago, but it's kind of great fun for me. And I've built 4 computers for my friends and wife, the 6600K system is my wife's I commncondeered for the benchmarks, and the i3 6100 system is one I was building for a buddy of mine last week.
The nature of my work means years spent in the third world, so I have struggled to have a decent rig overseas, that's the reason for the Origin 15 laptop which served me for about a year, though it does post some impressive (for a laptop CPU/GPU) benchmarks here @1080p, what isn't visible in the benchmarks is the thermal throttling that occurs within about 1-2 hours of use (depending on the game). This led me to build the i7 6700+GTX 1070 system in the RVZ01B case, specifically for Rome 2 with the Divide et Impera mod, as it was just too much for that laptop. The only issue with the small desktop was that it is still too big (barely) to bring as carry-on, so that meant wiping drives when I put it in checked baggage. So then I went back to the laptop course of action, resulting in the Asus G752VS OC, which has solid performance with that slightly underclocked GTX 1070 which hovers around 81C max temp, though the 6820HK is less impressive cooling wise as it regularly hits 100C, not uncommon for a laptop CPU, but not impressive at this point performance wise I'd say.
The i7 960 Alienware Aurora was my last desktop system purchased through Dell/Alienware in 2009, and I just got it back from loaning it to a buddy of mine for a couple of years; I had intended to run it with a 390X, but the POS Dell 875W PSU doesn't have enough amperage to run a 390X; so I had to resort to a cheap low draw option: the XFX RX 470 4GB single six pin, single fan. Looking at the performance of the i7 960 with this GPU which is inferior to the Asus STRIX RX 470 4GB dual pin, dual fan paired with that i3 6100, I would say it holds up very well, and for gaming would be fine to this day.
The i7 3930K Alienware ALX I bought off of eBay for $600,ran into the same issue with the PSU and tossed the Asus STRIX 1060 6GB in as a result.
The i7 4770K system was my first personally built PC in 2014 which I transferred and upgraded the AIO and motherboard into a 780T case (initial build I had it in an awful HAF-X case with a crappy MSI G-45 mobo). The 780Ti MATRIX cards are still quite potent, which I noticed when the GTX 980 came out as they often beat reference 980 SLI, and to this day one of them can beat my GTX 1060 STRIX card. It would be interesting to see how they would do with another 3GB of VRAM and another 300-400Mhz on the core in these benchmarks. I have a feeling their bus/bandwidth and shaders is what lets them keep up with the GTX 1060 6GB's higher clock speeds, considering that clock speed wise my 780Ti cards are only 1076 core/1215 boost/1778 memory vs my 1060 6GB's 1727 core/2113 boost/2100 memory.
The i7 5930K system I built as my main rig, I'm debating heavily on the 1080Tis as in a large number of games my 980Tis are within a few FPS of reference GTX 1080s in SLI @4K+ (The Division @4K, 4.3 FPS difference, Shadow of Mordor @5K, 1 FPS difference when using guru3D benchmark settings) and considering that in terms of FPS the difference between the 1080Ti and 1080 is only ~16 FPS across a 15 game average (Jarred Walton @pcgamer), most likely @7680x1440p that will drop in half, I'm estimating that in SLI I would only see maybe a 10-15 FPS gain over my 980Tis, unless it's a VRAM issue.
What really confused me while running these benchmarks was how SLI was all over the place, I mean 780Ti SLI should never beat 980Ti SLI, but it did @1080p Ultra (ran it several times to verify). Looking at the results it seems that SLI for the 980Tis does not kick in until certain settings: 1080p Max, 1080p Low, 1440p Max, 1440p Ultra, 2160p Max, 2160p Ultra. Unclear to me why 1080p Ultra, High, and Medium seem to not be using SLI for the 980Tis, because it certainly is for the 780Tis.
1080p Ultra was interesting to me as a single 780Ti was within 5.7 FPS of either of the 1070s. Again, I ran these several times, but results were always similar. Also @1080p ultra the i7 960 with the weaker RX 470 curbstomped the i3 6100/RX 470 STRIX, and was within 4.7 FPS of the i5 6600K/390X STRIX. I'm fairly certain that with a 390X STRIX the i7 960 would have beaten the i5 6600K in this preset.
For 7680x1440p I was forced to drop to the Ultra preset (ultra units) as completely maxed was breaching the 6GB of VRAM at that resolution. Same thing for the 780Tis @2160p Max, I couldn't run that preset with them due to the 3GB of VRAM not being enough. If they had 6GB of VRAM I'm certain they would have cleared the 30 FPS, possibly hitting 40 FPS, as you can see when I dropped down to Ultra @2160p the 780Ti by itself hit over 30 FPS while in SLI it got over 58 FPS. Ultra 2160p is also interesting as it is here that you see the cards still able to perform, but near Max capabilities; only 6 FPS difference between the 780Ti and the GTX 1070, and the GTX 1060 within about 10 FPS of the GTX 1070.
I do have some DX12 benchmarks done, but have not organized them yet. Also looking to see what happens when I try to run the benchmark at lower resolutions on my i5 dual core+GT 335M, probably won't even run, might even explode, not sure.
Re: TWC Total War Warhammer benchmark thread - last updated: 26.12.2016
Nice set of info Druout, lots of interesting data points there that provide some perspective. I will say though that honestly for finding upgrade information I found the data set from alQamar the most helpful because it showed comparative points with one factor being the same and then changing the other, followed by your 1x GPU vs SLI sets. Keeping the same 970 GPU and "upgrading" processors pretty much showed to me that I don't need to upgrade my CPU because going from a 2600k to a 7700k only gained 10fps (and since i'm on a 4770k i'd probably only gain half that). Considering a full base upgrade (proc+mobo+ddr4+cooler) would be like $950 it definitely isn't worth the gains. I'd love to see you do some tests with the top 3 I7s (5930k, 6700, and 4770k) with each of the top 3 GPUs to truly see what kinds of gains you get from upgrading each part, but naturally (and not counting SLI) that would take a while to swap the parts and get the 9+ test sets with all the different quality settings etc.
One thing i'm curious specifically about is the gains you're getting from OCing vs stock (especially on the 4770k since thats the processor I have :-P feel the self interest!). I found this forum because I was googling sites that have Warhammer in their test beds because right now it kinda is my primary game and i'm upgrading because I want to be able to fight 3 stack vs 3 stack with more than 15 fps but still on Ultra, and so I was looking around to see if it was worth upgrading my I7-4770k to a 7700k or the Ryzen 1800X (particularly was looking for Ryzen info). I found Warhammer in the testbed for the Ryzen review at http://www.gamersnexus.net/hwreviews...chmarks/page-7 and while they only had 1080p/High settings and not enough I7 comparisons I found it really interesting the gains that they were getting for OCing their testbeds (not to mention how poopy Ryzen is even with TW:W being an AMD backed game). It didn't even figure into my brain because I don't OC my stuff (I prioritize quiet PC over OCing gains and higher coolr noise) and looking at my CPU utilization when ingame and seeing it not even at 100% my brain never really registered that of course you'd get some gains because OCing is basically faster instruction handling instead of the "more" instruction handling that cpu utilization essentially is.
Anyways, i'll run some benchmarks myself to add to this data set, as I have a 1080TI preordered and am now going to look into those self contained water cooling systems to OC ($120 for 10+ fps gain from OCing vs 850 for 10+ fps for a ne platform) so I should come back eventually with some sets showing the upgrade differences off my current I7-4770k 3.5ghz + 980gtx setup vs 4770k OCd + 980 vs 4770k stock + 1080TI vs 4770k OCd + 1080ti.
Re: TWC Total War Warhammer benchmark thread - last updated: 26.12.2016
I think I posted this in the wrong thread initially. I ran a test on 4 of my systems Bloomfield, Haswell, Haswell-E, and Skylake using a total of 56 units on the Spire map (Carcassone in defense with 4 heroes, 8 Foot Squires, 3 Fire Archers, and 1 Trebuchet vs the Bloody Handz with 1 hero, 29 Goblin spear units, and 10 Skulkers). I placed Carcassonne in an outward L shaped defense with my only input being to keep the archers firing the entire time. I used the same static camera location for each 480 second FRAPS benchmark run.
Spoiler Alert, click show to read:
[IMG][/IMG]
Only the 5930K/1080Ti SLI rig puts up what one could call a playable average FPS, but even it spends a significant amount of time (roughly 220 of 480 seconds) below 30 FPS; however, unlike the other systems, frames are consistently above 20 FPS. I think it's safe to say that the quad core systems are within margin of error of each other and simply overwhelmed.
Last edited by Druout; August 10, 2017 at 05:46 AM.
I just bought a new PC and started TW Warhammer. I did run the integrated Benchmark and was astonished that there is a new one dwarves vs orcs and the old one orcs vs empire is gone. When did that happen. I did play a few weeks ago when the norsca DLC came out but I remeber there the benchmark was the old one.
Re: TWC Total War Warhammer benchmark thread - last updated: 26.12.2016
thanks for all the new posts, moderation merge and kind appreciation. Most of all a credit to Druout for his massive benchmarking session with so many variants. I pretty much know how much time he had to invest to get these here and on the legacy Attila thread. I am happy that this thread still helps out people to find answers on performance questions, hardware choices and settings.
Warhammer 2 is at the gates. Do you think we should make another thread or expect similar performance?
I heard about further graphics improvements and hope the new benchmark will be more practical than Warhammer 1's
Gesendet von meinem SM-G930F mit Tapatalk
Last edited by alQamar; September 20, 2017 at 12:42 AM.