Don't be angry just because you bothered to compose that post when it was completely unnecessary taking into account what DW posted in the first place, it's not worth it mate
Someone disagreeing with you makes them angry from your point of view?
I did take it into account, I didn't see the point in arguing it. I said not to believe what Youtube videos tell you, as they're poor sources of information. This refers to the claims that a budget CPU will turn those results completely on their heads.
And if you really want to be nitpicky, here: an actual look at the impact of CPU on AMD and Nvidia cards. This is Doom in OpenGL (not Vulkan), comparing i7 5930K vs i3 4360 (not an actual 4360, rather a CPU with disabled cores and reduced clocks).
http://cdn.mos.cms.futurecdn.net/7Ni9835hggpGXxfU26oBHQ-650-80.png
The 1080 falls to 83% of its performance on a stronger CPU. The Fury X falls to 88% of its performance. The weaker cards are unaffected.
Here with the game at Medium settings:
http://cdn.mos.cms.futurecdn.net/ZMiRTtXWCkss42W3DCu6iN-650-80.png
Here the 1080 falls to 94% of its pervious performance, the Fury X falls to 86%. This does a better job of showing the AMD limitation, yet the card is still performing well above 100fps.
The weaker cards are once again unaffected.
Reminder here that AMD performs poorly on OpenGL, even worse than they do on DX11. That's why they want to move on to DX12 and Vulkan. But despite this, the weaker CPUs do not somehow turn their cards into complete garbage.
They tested with 2 cores running at 3.7ghz. This is below what people buying a 480/1060 would even buy. Yes, I did read the full original post. Including the section saying that the CPU overhead is so large that it can make the 970 run better than the entire AMD lineup. If the 118$ cpu doesn't ruin the cards, I think nothing will. (Current 2-core 3.7ghz CPU: i3 6100)
If someone who buys the 480/1060 spends as much on their CPU as they did on the GPU, they will have a 4-core i5. Will this make the 970 faster than the Fury X? No, it won't.
I somehow thought I didn't need to make my original post longer to dispel something this silly. But hopefully now you are satisfied with the information presented.
I'm the only one who shared sources and actual tests. Yet people are accusing me of being inaccurate.
Unfortunately the cache size on the chips cannot be reduced, as far as I know.
So, not the best comparison between low/high end CPU's in gaming.
boundle (thoughts on cracking AITD) wrote:
i guess thouth if without a legit key the installation was rolling back we are all fucking then
Unfortunately the cache size on the chips cannot be reduced, as far as I know.
So, not the best comparison between low/high end CPU's in gaming.
That's 100% correct, that part won't be changed.
With that said, I think we can agree that if a 120$ CPU reduces a 350$ card by 14% (instead of 6%), a 210$ CPU with a 210$ GPU won't show the same discrepancy.
I'm not arguing against the idea that AMD has CPU overhead, only some of the ridiculous claims tied to it.
Well, nice to see the GPU pot stirred, we haven't had a feisty discussion on it in a while.
Kanint wrote:
I looked it up and only Shadow Warrior 2 ran better on the 970.
Youtube videos are generally not the best source of information.
Not sure how much time you spend on this topic, but the average user lecturing me on GPU and CPU usage, would be like me lecturing couleur on ethics and philosophy.
I mean, I spend pretty much 3-4 hours every single day researching GPU and CPU performance in games. I'm not pulling answers out of my ass because I own a 970. I bought the 970 because I researched it and it is was a perfect fit for an i5 system and the wide spectrum of games we get served today. And when it comes to triple A games in recent times, it's quite rare to have a game completely free of CPU limits in its entirety.
You reference journo articles with 16 thread cpu's paired with mid range cards.
hardwarecanucks review. let's look at the fallout scores. one of the most misleading benchmarks i've ever seen in pretty much every journo article.
No mention of the CPU overhead in downtown boston, and the results in no way reflect the reality, that a gtx970 will outdo the whole AMD line up there (until we get fully into higher resolutions where the GPU power becomes the limit for the 970) because the GPU plays a secondary role at mainstream resolutions in this game. Not a single mention of this in the hardware canuks review. so for me, that's misleading the consumer. you will not get a similar experience with a 1060 and a 480 in fallout 4. But they actually show the 480 having better minimum FPS.
I mean, show me an AMD setup that can do this? I've never seen one. Maybe a 16 thread cpu with ddr 4 4000 could get close to this?
i5 and a 970 (if i plugged in a 480 here, GPU usage would be at 20-30% in all likelihood)
I also see you say, an i5 won't make the 970 faster than a fury x. oh yes it will. it will do that in mafia 3, watchdogs 2, fallout 4, dishonored 2, as it will in every single CPU limited scenario. You are downplaying the CPU overhead. We are talking 20-30 fps differences in certain titles. I mean I could post videos all day.
i5 4670k 970
390 i7 4770k (might as well call the 390 "any enthusiast level AMD card" here, because it doesn't matter, the GPU is twiddling it's thumbs here waiting for grandma CPU overhead to knit the sweater)
dishonored 2 next, according to guru3d simple graphs, the 480 beats the 970.
notice his fluctuations between 40's and 50's all the time, rarely pegged at 60?
At those same settings, in the same exact area, I never dip below 55. i5 and a 970. The difference in min fps and framerate stability is very noticeable. And yet journos report the 480 doing better. yeah, sure, in the ideal scenario of being GPU limited the faster card pulls ahead. but gaming isn't ideal scenarios.
And what makes it worse? lowering settings doesn't help. On the other hand you got AMD favoring games like hitman, where amd cards clearly take the cake. but the difference is, you can lower settings and get more FPS with NV. because the case here is that the game can make use of AMD's superior hardware specs to pull ahead. It won't prevent nvidia from reaching those frames by lowering settings.
you cannot do that in a CPU overhead scenario that AMD often faces. That's why I consider it budget at this point it time. You get a more powerful card, more VRAM, but when you cast a wider net of different rig setups and gaming scenarios, you will start seeing why you got those things for the same price.
Now, does someone wanna gamble with that, and hope the games they play won't be affected? that's fine. Does someone know specifically what they play, and know AMD will serve them better in those titles? fine as well. Heck, even some brand loyalty is fine by me. support the company you like better if that pleases you.
But let's not forget the current maker share situation, and what caused it. The fact that you often need an i7 system for same price bracket AMD cards to match or exceed NV cards with i5 systems, and the fact that the CPU overhead will often leave you with 50% gpu usage on your GPU with superior technical specs. And again we are talking about real gameplay that we do, not 1 minute benches in specific spots done by review sites who have deadlines and sponsors.
And this translates all the way down to the budget line. pair an i3 with an nv budget card, and you will be seeing higher FPS than that same CPU with an AMD card. the same AMD card that would blow that NV card out of the water on a 16 thread CPU.
If we allow personal anecdotes:
Out of all the games you've mentioned so far, the only one I haven't tried personally is Dishonored 2. Out of all the ones I've played, the only one where I found myself limited by my CPU was indeed Fallout 4, probably the worst game there is for CPU usage out there. I agree this is a very poor game to use for benchmarking purposes. This IS a game where I felt I would get slightly better performance (especially in Boston) with Nvidia.
Regarding Watch Dogs 2: you're comparing someone running with Temporal Filtering On vs someone who isn't (even when he uses the low preset, it's not on, none of the preset turn on Temporal Filtering, in fact every preset turns it off!). In the WD2 thread you helpfully demonstrated the difference that ram speed makes. So I think you know the difference between someone running with Temporal On vs someone running with it Off + 2133 ram.
I'm running Watch Dogs 2, i5 4690k R9 390 ram at 2400mhz. Even with the low preset, my GPU usage is maxed at 100%. With my resolution at 2560x1440, I used the Low preset, reduced my FOV back to the minimum, and turned Temporal On. This is the best case scenario to see if my CPU will limit me. My CPU usage got high, peaking at 91%, but my GPU remained at 100%. At 1440p with the lowest settings possible, I'm at 84fps and limited by my GPU, not CPU.
Using my normal settings, FPS currently at 57, CPU goes from 65 to 72, GPU capped 100.
I'm using the latest drivers released today, and some people claim that they include some further DX11 overhead improvements, so maybe that helped. But I've been playing WD2 since release, and I've always been limited by my GPU when I looked at Afterburner.
When I play games with very, very poor CPU usage (such as Fallout 4, a poor game to benchmark in any case, or MMOs), that's where I see my CPU limiting me. But that's 1 game out of those you've listed, and I don't hold the Creation Engine in high regards.
The 4690k is slower/older than the 4770k/6500 used in those videos, but it's not limiting the GPU outside of FO4.
Well 'rented' myself a gainward gs 1080 GTX and if i don't bring it back within 2 weeks i made it a 'buy'. Cuz gaming on a HD 390 from ATI isn't sufficient for gaming at 4K with 60 fps actually it's 30 fps on average but the immersion is so much better then gaming at 1440p let alone 1080.
The GS version seems really to be a golden sample cuz out of the box i'm getting 2000 mhz boost speed and when clocking with 100 mhz i'm getting 2150 mhz and i might get more when overvolting but you can't game with it and i only overclock when i can actually use it when gaming. But when handling the fan myself manually it stays around 2000 mhz i noticed when having the fan at 75%.
I'm getting twice the speed of the 390 in most games some even a bit more, some game 1.5x and some games only 1.25x.
But i'm waiting for the 13th when hopefully ATI will show some more about their upcoming cards and what they actually deliver, the 490(x) fe and some benches of them which gives us an indication which power they will have. I'm very looking forward to it cuz what i'm doing now is just cursing in church as they say here in Holland But it's just a lend ( i think ) Well there's another downside, i bought this card as well as i noticed euro was falling against the dollar and i bought a 'demo' card and the same card is now 120 euros more expensive in the same shop.
I'm always bringing myself into troubles when Christmas comes around and i have enough money to spend for a high-end card, always in october and november when there are a lot of new AA games released and you can't play them like you want. Of course in good coorparation with the GPU designers.
Regarding Watch Dogs 2: you're comparing someone running with Temporal Filtering On vs someone who isn't (even when he uses the low preset, it's not on, none of the preset turn on Temporal Filtering, in fact every preset turns it off!). In the WD2 thread you helpfully demonstrated the difference that ram speed makes. So I think you know the difference between someone running with Temporal On vs someone running with it Off + 2133 ram.
I'm running Watch Dogs 2, i5 4690k R9 390 ram at 2400mhz. Even with the low preset, my GPU usage is maxed at 100%. With my resolution at 2560x1440, I used the Low preset, reduced my FOV back to the minimum, and turned Temporal On. This is the best case scenario to see if my CPU will limit me. My CPU usage got high, peaking at 91%, but my GPU remained at 100%. At 1440p with the lowest settings possible, I'm at 84fps and limited by my GPU, not CPU.
Using my normal settings, FPS currently at 57, CPU goes from 65 to 72, GPU capped 100.
The watchdoge vids were more to show the AMD system not gaining fps with lower settings. Great to hear you are seeing good CPU and GPU utilization, would be even greater if you had video evidence of it, to put more data on the internet with regards to AMD gear, as I said, it's often difficult to find for many games.
i've not been able to find any watchdogs 2 AMD footage with consistent FPS during fast speed city driving and people are reporting that changing settings doesn't improve their FPS.
And if we follow journalists, all is well for both camps in every game, somehow, despite internet wide evidence of otherwise
Anyways, for most of the games I mention, I have at least some footage of on my channel, and i'm always open to more tests and comparisons, so that we present unbiased data about GPU and CPU performance, something journalists are failing to do in more cases than not.
The watchdoge vids were more to show the AMD system not gaining fps with lower settings. Great to hear you are seeing good CPU and GPU utilization, would be even greater if you had video evidence of it, to put more data on the internet with regards to AMD gear, as I said, it's often difficult to find for many games.
i've not been able to find any watchdogs 2 AMD footage with consistent FPS during fast speed city driving and people are reporting that changing settings doesn't improve their FPS.
And if we follow journalists, all is well for both camps in every game, somehow, despite internet wide evidence of otherwise
Anyways, for most of the games I mention, I have at least some footage of on my channel, and i'm always open to more tests and comparisons, so that we present unbiased data about GPU and CPU performance, something journalists are failing to do in more cases than not.
I rarely record, so forgive me for the horrible quality.
I kept to the lowest settings, same as before, still at 2560x1440. Running on overclocked 390 (by around 12%) and overclocked i5 4690k (4.5ghz).
I picked the large straight street in the middle of San Francisco (the only thing 'straight' in San Fran) and simply drove one way then came back. At the end I stayed in one spot with a high CPU usage. Overall, as you say CPU usage is pretty damn high, it easily got over 95% in many spots and both my CPU and GPU were maxed out. But my FPS kept between 70 and 80, didn't fall down under 70, nothing major from the CPU getting that high. After activating the nitro there was a point where GPU usage got down to 95% while FPS remained at 70, that's the worst that happened.
The overclocked 390 is faster than the stock 480 and I'm running a CPU released in mid-2014. If people can't get consistent FPS while driving, I would look elsewhere for their issue. I can see why you look for feedback on youtube, but this is why I personally don't put much trust on those videos: it's easy for the user to be mistaken as to why they get poor performance. It's possible the issue is the user, not their hardware.
Take it as grain of salt, but looks like AMD´s Pascal killer will be level of aftermarket GTX 1080 (in a vulkan title), and Nvidia has 1080Ti in their sleeve . If this is the case, i wonder how this will be priced. doesnt look that impressive after this kind of delay. Might be shit drivers and bullshit numbers, but hey what we know until "real" benchmarks surfaces.
The watchdoge vids were more to show the AMD system not gaining fps with lower settings. Great to hear you are seeing good CPU and GPU utilization, would be even greater if you had video evidence of it, to put more data on the internet with regards to AMD gear, as I said, it's often difficult to find for many games.
i've not been able to find any watchdogs 2 AMD footage with consistent FPS during fast speed city driving and people are reporting that changing settings doesn't improve their FPS.
And if we follow journalists, all is well for both camps in every game, somehow, despite internet wide evidence of otherwise
Anyways, for most of the games I mention, I have at least some footage of on my channel, and i'm always open to more tests and comparisons, so that we present unbiased data about GPU and CPU performance, something journalists are failing to do in more cases than not.
I rarely record, so forgive me for the horrible quality.
I kept to the lowest settings, same as before, still at 2560x1440. Running on overclocked 390 (by around 12%) and overclocked i5 4690k (4.5ghz).
I picked the large straight street in the middle of San Francisco (the only thing 'straight' in San Fran) and simply drove one way then came back. At the end I stayed in one spot with a high CPU usage. Overall, as you say CPU usage is pretty damn high, it easily got over 95% in many spots and both my CPU and GPU were maxed out. But my FPS kept between 70 and 80, didn't fall down under 70, nothing major from the CPU getting that high. After activating the nitro there was a point where GPU usage got down to 95% while FPS remained at 70, that's the worst that happened.
The overclocked 390 is faster than the stock 480 and I'm running a CPU released in mid-2014. If people can't get consistent FPS while driving, I would look elsewhere for their issue. I can see why you look for feedback on youtube, but this is why I personally don't put much trust on those videos: it's easy for the user to be mistaken as to why they get poor performance. It's possible the issue is the user, not their hardware.
that looks good yeah. Do you get the same GPU usage even at higher settings? looks like the RAM speed is likely what is holding back most of the nubzors over the internet and likely has a slightly higher effect on AMD users due to the overhead. that's why I wish I had a profitable journo site under my control.
ATM, we get the both extreme ends most of the time. journos that test every gpu for 2 min, on a super computer, or incompetent users who pair up 1080's with AMD FX cpu's
I literally had to post a vid, where I alt tab, and open up gpu-z and cpu-z, and type stuff in notepad, because the average internet dolt thought I was making stuff up when saying I get 60+fps when driving. and still, the video has more downvotes than upvotes.
Will buy a 1070 myself too. But not now. As the current price isn't worth the performance gain to me (money is not the problem, but i have this attitude).
Enthoo Evolv ATX TG // Asus Prime x370 // Ryzen 1700 // Gainward GTX 1080 // 16GB DDR4-3200
nah bob, the good old 770 dotched the bullet aka 970 but on purpose, always skipped one generation of gfx cards. though the 1070 now is the first one since years i buy directly and not a little bit cheaper used. payed 438 euros, already sold my 770 for 100 euros inkl shipping. and if i sell the wd2 code it would end up with spending 300 euros (ok a bit more, ordered ram too to get from 8 to 16gb; that system will last again 2-3 years without problems). thats ok, but i most likely will send pumpy the code for the xmas giveaway
and a little questions to the 1070 owners
reaching 2ghz boost with ~0,95 volt should be possible without problems right? what about the memory? worth to boost it up too?
edit: ah i love the german post,ordered it yesterday @ noon, just got delivered.cant wait to put it into my system later
replaced the old stuff -> monitors not responding. cmos clear. still not responding. building back the old gfx card -> still no picture. old ram -> same. FUCK MY LIFE! seems my board or pci express slot died! wtf?
was at a local store to check different gfx cards to psus too. nothing works...
na normally not, im not using those useless wristbands but i touched a radiator before. thats always fine.
only thing i can imagine -> i damaged the pci express port. which i doubt cause i found a pci gfx card which doesnt work as well (or the old card is damaged too, dunno. its 20 years old or the board doesnt even accepts pci gfx cards)
other stuff: i disabled the xmp before switching components (cause the old ram was 1,65v and the new one is 1,5). after hitting save changes and quit uefi the pc shut off completely (which might be normal when changing the ram timings etc) but hasnt switched on alone. havent tried it manual cause i didnt need to boot again befor switching the components. so maaaaaybe it killed itself when changing the ram settings, dunno. 1.65v isnt standard, 1,5v is...
just tried it again (with the new ram and old gfx card) -> it shuts itself off after ~30-40secs and switches on again a few secs later. might be a sign of problem with the ram banks cause the board tries different things on the ram straps (atleast my old board for the core2quad behaved similar when the ram made issues)
im very pissed atm (and needed to spend another 80 euros on express delivery via ups for a new board for tomorrow, got an gigabyte z97 hd3 cheap for ~80 euros including ups express delivery stuff). hope it will work, but i will loose my full 3,7ghz boost on all cores with that board
na normally not, im not using those useless wristbands but i touched a radiator before. thats always fine.
only thing i can imagine -> i damaged the pci express port. which i doubt cause i found a pci gfx card which doesnt work as well (or the old card is damaged too, dunno. its 20 years old or the board doesnt even accepts pci gfx cards)
other stuff: i disabled the xmp before switching components (cause the old ram was 1,65v and the new one is 1,5). after hitting save changes and quit uefi the pc shut off completely (which might be normal when changing the ram timings etc) but hasnt switched on alone. havent tried it manual cause i didnt need to boot again befor switching the components. so maaaaaybe it killed itself when changing the ram settings, dunno. 1.65v isnt standard, 1,5v is...
just tried it again (with the new ram and old gfx card) -> it shuts itself off after ~30-40secs and switches on again a few secs later. might be a sign of problem with the ram banks cause the board tries different things on the ram straps (atleast my old board for the core2quad behaved similar when the ram made issues)
im very pissed atm (and needed to spend another 80 euros on express delivery via ups for a new board for tomorrow, got an gigabyte z97 hd3 cheap for ~80 euros including ups express delivery stuff). hope it will work, but i will loose my full 3,7ghz boost on all cores with that board
Have you tried with single RAM stick (old and new) and do you have integrated GPU in your CPU?, also do you have MemOK! button in your mobo?.
I hope its the board. If not i will return the New board and the ram i ordered. Try not not cry,lay down,start crying. And will spend money on an new board,skylake i7 and ddr4
Signature/Avatar nuking: none (can be changed in your profile)
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
But it's just a lend ( i think ) Well there's another downside, i bought this card as well as i noticed euro was falling against the dollar and i bought a 'demo' card and the same card is now 120 euros more expensive in the same shop.
