Gadgetory

hey Ron i'm steve from gamers nexus donna and this is another ask GN video this is our fourth episode i believe of ask GN and i'm pretty excited with how they've gone so far actually because the questions that you all proposed have actually sort of inspired us in a few ways to do some new articles new tests things like that and it's gotten us to think about hardware in sometimes ways that we haven't before or we haven't done it a long time so please keep the questions coming a post them below on this video i check the most recent ask GN video first before filming the next one and i normally film only one day in advance so I'm trying to do these every monday at this point so today the first question we have is a pretty simple one it's from sorry about the pronunciation here nishant chikara who says any news on amd's NCP you launched so i just posted an article about this and the AMD Zen architecture for those who don't know is Andy's next architecture following the bulldozer piledriver steamroller lineup which we are presently on so the current kaveri apu czar using steamroller modules in them which used to be called bulldozer modules still kind of correct but they're upgraded and the next architecture from andy is the zen architecture and that has so far a lot of hype behind it there were a couple rumors from motherboard manufacturers who were unnamed as told to i believe it was digit times maybe i think it was digit times said that unnamed motherboard sources predicted a 40-percent performance gain with zen over the current architecture that's a very big claim so mountain of salt with that it's not verified but that's sort of what we know so far in terms of the launch date there's no hard launch date as of yet the current rumor is fourth quarter 2016 that's about a year from now and that's unconfirmed it could be later could be earlier but I really sincerely don't think it will be it's we're looking at fourth quarter 2016 or later at this point maybe third quarter but I don't think that's going to be the case dark drifter best sub $100 cpu for gaming so depending on how sub $100 you mean there are a couple of options here the most immediately one I would point toward would be something like the 41 30 or 40 160 I three CPUs those are priced at one hundred twenty dollars so it's not technically sub 100 but it's very close and the twenty dollars gets you a lot of performance over the actual sub 100 dollar CPUs which would be the g3 258 pentium and the 860 k soon to be replaced by the ATK in theory that ever comes out so those would be the three to look at if you can afford the i3 at around 120 ish dollars it's it's very good we found it to be basically unless you're using a high on GPU that's going to bottleneck on the cpu we found it to effectively as good as most gamers will need unless you're really starting to push thread intensive games and there's not a lot of those out there so if you are doing thread intensive games you probably already know that you are but those would be things like the CryEngine for instance has the ability to use all the threads it doesn't always do that but it's threaded for up to eight spawns so it can spawn eight threads one of those is the physics processing one is the game logic and then you've got rendering and all that other stuff in there as well so CryEngine games have the ability to use more threads but for the most part you'll be just fine on an i3 that said a lot of people will be just fine on an 860 k as well and even a g3 258 so unless you're playing like GTA 5 which we talked about in a similar discussion previously then the 860 k and the g3 258 are fine CPUs at the sub 100 dollar price point the g3 258 I think is maybe five dollars cheaper than the 860 k which is a $75 CPU and the 860 k the only reason I generally recommended over the g3 258 which is still very good penny mcp you is because the pentium the g3 258 is it has trouble sometimes with heavily multi-threaded games like GTA 5 it will push a high average FPS 75 the case of the most recent test I did with a 980 TI it's pretty darn good for normal settings on GTA 5 with a pentium dual-core cpu but its point one percent low numbers are not very good and i think i talked about what those numbers mean in episode 2 of ask GN so check that out if you don't know but the point 1 percent low is on the pentium were like for FPS something around the head range and the 860 k even though i had a lower average in the 60s average FPS the point 1 percent was much higher to the point where is actually more playable so that is what you're looking at for the sub 100 dollar range but if you can afford the twenty dollars i would push into the i3 area otherwise 860 k would probably be one of my main go-to sg3 258 if you're not playing heavily multi-threaded games and it overclocks really well so throw that in there too next question how r thompson says hey Steve do you think the I 759 30k is that all worth it over the I 758 20k to run to 980 TI video card debt by 16 by 16 or by sick by 16 by eight excuse me for 4k gaming so the 59 30k in the 58th 20k those are two x 99 CP is for those who don't know will run through the specs right quickly the more expensive one would be the 59 30k it's in the 500 to 600 dollar range the other one the 58 20k is in the 309 de la range at last check on newegg so those are the two price points for x 99 without going into the thousand-dollar really high-end cpu area both of them are six core cpu is both of them are clocked about the same i think it's three point three gigahertz where's is 3.5 somewhere in that range and so you're getting a pretty small clock increase but what's actually the thing you're paying for is the PCIe Lane count increase the 59 30k as 40 lanes and the 5820 it has 28 lane so that's what Homer is asking about here is is the extra Lane count worth it to run two video cards and x 16 x 16 versus being forced down to x 16 x 8 or x 16 x 8 depending on how you want to say it so the the lane assignment here is definitely a thing to research and be concerned about as you're doing it's a good thing to ask about but in general for the current architecture GPUs they're not pushing on a throughput to actually saturate that by 16 PCIe 3-pointer face you're really not going to have that much to worry about at x 16 x 8 vs x 16 x 16 the difference is negligible and this is something that we're going to test and publish soon i have tested it but not comprehensively enough to publish charts just yet it's something we're going to do soon other websites have done it in the past we're just going to refresh it but if you're curious you can look at past benchmarks that have done this exact scenario different CPUs but exact scenario otherwise and the general idea is that because the throughput of pcie 3.0 is so high just with one lane you with eight lanes with 16 lanes you're going to see effectively identical performance even an SLI now that said you're instantly consuming all of your lanes so if you run with the 28 lane 5820 k you're not going to have really any lanes left over for other pci or expansion devices so if you're running we're hoping to run an m2 SSD the m.2 ssds will generally the good ones will want four lanes dedicated to them or at least two so do keep that in mind that's where you start losing performance is when you're trying to throw in other PCIe devices or if it's not a PCIe device keep in mind that other interfaces will still use PCIe lanes so an m dot to interface on the motherboard will use PCIe lands from the chipset or the CPU or somewhere in the system and that's true for some other devices as well to answer the question very directly no is it is not worth it strictly for that scenario but it is worth it if you want the allowance for another device so hopefully that makes sense please let me know if anyone has questions about the subject and the last question for today is from Josh orenburg as another question does Andy have physics equivalent what's the situation on the AMD side this is another good question and this is one that's been around for quite a while now physics got its dawn when I gia was purchased by and video or its patents were purchased by Nvidia and that happened a number of years ago so for those who don't remember the company they made physics processors so they basically took what is now an extra GPU they made a physics coprocessor they put it on a card you put that on your pc i interface PCI Express interface and then that handled all of your game physics processing they were actually pretty popular and they were very high-performing toward the end of their tenure so Nvidia bought that technology and turn it into physics which is a solution that now lives on the GPUs and in the drivers of the video cards that they produce physics will technically run on AMD devices really not always the best it doesn't always work either so it just depends on the game and how intensive it is but does Andy have a physics equivalent no they do not Andy is very good at things like opencl and compute actually better than a video in some ways but that doesn't really show when a lot of games because of how games are built and that's something I discussed in the fury x review check that out if you're curious but Andy's where you get an open CL and the also is basically optimizing for things like the havoc engine which is a physics engine that game developers use is pretty popular one actually and build into their games to handle all the physics computations so when you're using something like havoc some of that will be it's basically software accelerated physics so it does all the processing and it's all coded through an extra engine that is available to the game developer so it's simplified for them by the havoc team but then the GPU depending on how its programmed is capable of taking some of that computation and running it itself rather than sending it through the CPU which is a sequential processor cpu is not very good unless there's some exceptions to modern CPUs but in general they're not very good at processing things that live much better on parallel architectures like physics does so Andy basically takes their opencl advantage where it's applicable they will use that advantage to do all the processing on the GPU without building their own physics equivalent and it's not always possible so in the cases where it's not they rely on things like the havoc engine and other sauce or accelerated engines or hardware accelerated software solutions depending on how you want to word it and they'll rely on those to do the the work of pushing the numbers around to different devices there is no physics equivalent out to answer the question very plainly so that is all for this ask GN episode I am currently on the way to another meeting in another state so traveling again but we'll have a lot more videos once back and that will be on tuesday or wednesday so back pretty shortly here this time not gone very long and check out the website for more articles recently both wrote a big one about AMD and the current situation it's got a lot of financial information pretty heavy on the number side it's also got a lot of information just about recent launches future launches like the fiji and zen stuff that's going on so check that out if you're curious about how the industry is going and then of course the witcher 3 video we posted recently is pretty cool too i think because talked about the financials of the game industry and what a game costs so that's all for this time please post more questions below if you have them I really appreciate them they're pretty fun to work on and give us a few article ideas as well if you like this content as always hit the patreon link in the post roll video we've gone up another one or two patreon supporters since the last time that I mentioned it so big thanks to all of you who have contributed thus far and we hope more of you will jump on board and support the site so check that out and I will see you all next time