Its modern, id say its low end tho. 770 would be mid range and the 780 high. Titan is enthusiast level same as the new AMD R9 290x.
A 760 is mid to high end, low end is a 6670 pr 7730. Anyway PA doesn't need much to run well. Runs lovely on my laptop which has a GT 420M (96 fermi cores)
I have an i7-3770K and a GTX 770 4gb. PA doesn't seem to push any of this at all, and I with it would, it would certainly reduce lag partially caused because the game data is streaming to my PC.
GTX760 is the low end of the current generation. GPU's generally dont decrease in price unles they are 2 or more years out of date.
Lol! It is categorically NOT low end. PC gamers are so elitist. The GTX 760 is based on the GK104 gpu (1 asic down from Nvidia's top). True it has 2 SMX units disabled, however it still has a reasonably wide 256 bit memory bus, 1152 shaders and a full 32 ROPS. This thing is capable of MAXIMUM settings in the majority of modern games at 1080p or higher. How is that 'low end'? (http://www.tomshardware.com/reviews/geforce-gtx-760-review-gk104,3542-10.html for a review on the card- this is up there with the Radeon 7950 Boost) It is also very much in the middle of Nvidia's current line up: {Ultra high end GK110} 1600p gaming / multi screen Titan GTX 780 {high end GK104} 1600p or 1440p gaming GTX 770 GTX 760 {mid range GK106 / GK107} 1080p gaming GTX 650 TI Boost GTX 650 TI GTX 650 {low end GK107} 720p gaming GT 640 GT 630 {entry GF108 or GF119 Fermi} Plants vs Zombies GT 620 and GT 610. All these are based on various Kepler asics (with the exception of the 620 / 610). Yes the lower parts are all 6XX series rather than 7XX parts- but the 7 series are still all based on Kepler chips, and are all re-brands- these are all current gen components. Even the Titan and 780 are based off the GK110 that was released at the same time as the 600's (however they only supplied these on Quadro products initially, probably due to limited yield of the large die). The truth is the only 'new gen' cards to be released in the last year have been AMD's 7790 and the just released R9-290X and these are likely to be the last all new products until the middle of next year when 20nm manufacturing is ramped up (Nvidia will be releasing the 780ti which I'm sure will be superb but it won't be 'new' on the basis its another GK110 based part). This card also costs over £200- again for one component I hardly call that 'low end'. The most I've ever paid for a graphics card was £130 for my current GTX 560 (got it just after the 6XX cards came out hence the low price). This was a comfortable mid range card at the time (and to be honest is still more than capable for current games). I guess it is a matter of perspective- however I think I'm fairly safe in saying that the majority of people playing PA have lower spec graphics cards than a GTX 760.
what do you see as the current generation? to me, the 600 series is not the current generation... the gtx760 IS the low end of the current generation.
Well I view gpu generation based on the chip they use rather than model numbers. Its quite usual for both Nvidia and AMD to keep the 'older' branded cards for the lower end if there is no more performance to be had. So for example, going back a little the GTX 4XX was the same generation as the GTX 5XX because they are both built on the Nvidia's 'Fermi' GPU range. Nvidia did eventually rebrand all the 4's to 5's (going all the way down to at GT510) however those were released along time after the first 'new' GTX580. Similarly the 'new' AMD R7 / R9 gpus are the same generation as the HD7XXX series (although they have slotted in a few new cards: the R7-240 and R7-250 are new to desktop as those GPU's come from their laptop range to replace the ageing HD5000 based low end stuff, and the R9-290 / 290X are also based on new silicon). So from my point of view the 'new' GTX 760 is still based on the same silicon as the 6XX cards (in fact its on the GK104 GPU which is the SAME GPU as was in the previously 'top end' GTX 680). You are technically correct that it is the lowest performance card with a GTX 7 branding- but until (or if) they flesh out the lower end cards with 7XX branded parts, then the 6XX cards are still current in those segments.
In Galactic War, the server seems to be local - and it seems unable to use multiple cores. At around 5000 units in the game, one of my cpu cores was at 100% and the game was practicaly frozen. UI still working fine, but units not moving anymore. Had to quit it.
While the CPU cannot be used to help in rendering in the sense that the CPU would do the tasks normally done on the GPU, the CPU definitely can be used to reduce the workload of the GPU by doing some additional calculations to reduce the amount of work given to the GPU while still getting the same visual results. For example, the CPU probably already doesnt tell the GPU to render things which are nowhere to be seen on the screen. The CPU is used to reduce the GPU workload. There are more things to be done. For example, it might be that units are rendered even if they are covered by their icons (and they most often are), when it is completely unnecessary to do so. I dont know if they have an optimization for that case, but that would be one place where tiny amount of CPU power can be used to greatly reduce GPU workload. A large part of what the CPU can do when it comes to rendering is figuring out what actually needs to be rendered and at what level of detail.
The server still wasn't local, so it was mostly the server failing to respond properly. This shouldn't slow down the local client, but sometimes it does since the client gets stuck waiting for the server if the network part is getting really slow. Client isn't properly multithreaded yet. Server is currently getting threaded in the current PTE builds (foremost: dedicated network/chronocam thread), so client lags induced by server lags should no longer occur.
yes, exactly, wtf. look, I know the game drops to 11fps and your computer is barely breaking a sweat. That is because, your computer isn't running the entire Planetside2 Connery Server on it when it plays, it runs your window and only renders your vision ingame and your actions. The Sony server does the x100 load. same here. Your computer is literally playing the equivalent to a web browser game. The server is doing the work. The server will never be faster or slower based on how strong your machine is. now, what happens when you play local games or host? Luckily, the server takes a single CPU core just to get 11fps, so the 33% your using will still be free to play on, and for hosting the gpu does NOTHING so that your gpu plays the game on monitor while the CPU actually hosts the game for you to play. if you have very low gpu and CPU or integrated, you may be really slow local, you might not be able to local and play on same computer, the server actually takes enough a load off to allow you to play at all. if you have dedicated gpu and a dual core CPU, I predict just as good performance as current Uber servers or better. Godlike comps will probably be reach the 60 fps mark by this logic, no? this poll isn't opinion. Its fact. There is a correct answer. And its no. No matter if you think antibiotics will help the flu, the fact is no it won't help your game get more fps. Its like a poll, what color is the sky on a bright sunny afternoon, purple, orange, brown, or blue? Well, not really an opinion...
You obviously haven't done much looking into it. A GTX Titan has 6GB of Memory. Two GTX 770s in an SLI will not only have more memory, but equal clock speeds, for $200 less. (As well as less power usage.) As for it not being enough bang for your buck? You can improve FPS by over 70% just by adding a second card, instead of spending more to just upgrade.(PA doesn't have an SLI profile...yet, so there's no reason to rush for it.) As for voltages and overclocking, adding a second card isnt going to mess with it. It really wont. Connecting to card together in SLI/Crossfire is not going to modify and of the voltages that the manufacturer or the user set them to, or any of the corresponding voltages.
Memory of dual-GPU cards and SLI/CFx is not shared. So it doesn't no mater if you have 1 GPU with 2GB GDDR5 or 4 GPUs with 2GB each; they all will only able use 2GB in total. And if one of GPUs only have 1GB then each card will only able to use 1GB. Also keep in mind that two GPUs have twice higher power consumption and require more powerful PSU, they also produce more heat and which mean more noise or more expensive cooling system. I already tried number of multi-GPU configurations and now I'll always stick to single GPU. PA isn't only game that not using SLI and there also some other limitations that SLI adds on the table. Also I mostly forgot about that, but consumer-grade CPUs only have 16 lines of PCIe more than two GPUs just won't work efficiently and not all motherboards can even handle two properly. Though I suppose with with PA client-server architecture game doesn't even need SLI because it's should be technically possible to create multiple GL contexts per-GPU or even run multiple client instances where only one instance "talk with server" and share data with another. This won't work for FPS game, but for RTS like PA it's should be feasible.
That's not an issue. PCIe 3.0 8x is still fast enough for any CG application, not to mention that the GPUs are still interconnected via a separate PCIe bridge. Actually even PCIe 3.0 4x is enough to fire both AMDs and Nvidias top models. Even though SLI/Crossfire comes with a huge penalty: It scales terribly with the complex layout of modern render pipelines. If you are very lucky, then the graphics engine will support split frame rendering which actually improves responsibility by reducing the time required to render each frame. In that case, SLI/Crossfire can actually replace a more expensive single GPU. If not, the GPUs will fall back to alternate frame rendering which does not only come with the usual problems like micro stutters, but also does not improve frame times. And if things go really wrong... Then you end up with no SLI support at all, or even broken SLI support where the framerate drops even further.
Interconnection only used in few rare use cases and don't help with CPU bottleneck at all. Also if I remember correctly it's now deprecated on new GPUs. Ok I agree about PCIe 3.0, but there is a lot of people with Sandy Bridge CPUs and older that doesn't support PCIe 3.0, many motherboards also don't support it. E.g most of players with FX AMD CPUs don't have it's too and on AMD APUs the CPU cores will be bottleneck SLI/CFx performance. Considering how small progress even Intel CPU doing there is completely no reason to buy new CPU instead of FX/i7 anytime soon. So setting up just SLI/CFx isn't as cheap as it's looks like and GPU upgrade usually cheaper. Other important note why new GPU is cheaper: there is tons of people who agree to buy used GPU, so when I sell old GPU I getting something like 30-60% for new one already while with SLI/CFx upgrade it's will be harder to sell two outdated GPUs just because low-end/middle-end GPUs become cheaper faster..
Sandy Bridge provides PCIe 2.0 and 8x bandwidth is sufficient. More than just that, it's still 4GB/second. Considering that you are mostly updating vertices only, that's quite a lot. Even with virtual texturing and offloading textures into main RAM, in worst case it's not even a single second until the GPUs VRAM is filled. PCIe bandwidth only becomes an issue for GPGPU applications where you are performing efficient reduce operations on large chunks from main memory.
