Gadgetory

fractal design cases have inspired modders all over the world who have built some amazing systems like this Darkseid themed case by George priscilla's showcasing the spacious internals in the define s or metallic acid a mini ITX system by Justin Olson featuring a white black and red color scheme and a super clean layout in they define nano s there are a ton more awesome build like these on fractal designs modding series page so check it out via the sponsor link in this video's description and get inspired for your next project how's it going guys and welcome to my thread Ripper review Andy's first ever high-end desktop platform launches today and it's been talked about for months so you're probably already familiar with the specs on screen right now today I'm going to be comparing the 16 core thread refer 1950 X to the 10 core Intel 7900 X because they both cost $1000 and because Intel does not yet have a 16 core CPU in their X 299 platform that's available so I'm working with what I got let's get right into the testing setup so so for the threader per system here I have the AMD risin thread refer in 1950 X 16 cores 32 thread CPU I also got the Asus X 399 zetas extreme motherboard it's running the latest UEFI version provided by eight suits that's 0 for 0 1 & 4 CPU cooling I had the Corsair H 100 IV to 240 millimeter all-in-one liquid cooler now note on CPU cooling I used DTR for bracket that came with a CPU because I wanted to see how middle-of-the-road 240 millimeter AC tech based cooler would work with this configuration since I'm guessing that's what's the most common and AMD does seem to be kind of banking on the fact that people will have these somewhat available to them I also used a little bit extra thermal paste to get as much contact between the block on the cooler and the extra-large threader for heat spreader as possible to try to get maximum coverage over all four of those CPU dies even though only two out of the four are actually functional for GPU though I have the asus strix GTX 1080p i/o see it's running at the out of the box manufacturer overclocked speeds for memory have for 8 gig sticks of GCL Trent and DVD are for 3200 memory and quad channel the memory settings were manually keyed into the UEFI for both systems speeds are at 3200 timings are 14 14 14 34 and voltages at 1.35 finally for storage as the Windows 10 operating system on a Samsung 9 6 crow 512 GB nvme SSD and then games the video files are on an external SanDisk Ultra - 1 terabyte SATA SSD which is connected via USB 3.0 this system is powered by real as well tachyon thousand watt 80 + Platinum power supply and as you can probably see it's an open testbed for the skylark X system a very much the same configuration of course different CPU the Intel Core I 970 900 x10 core 20 thread CPU for the motherboard I have MSI X 299 gaming m7 ACK and then for storage I have Windows 10 on the Toshiba Oct Rd 400 512 gigabytes nvme SSD then it's all powered by the gigabyte extreme XP 1200 M 1200 watt 80 + Platinum power supply beyond that everything is exactly the same between the system's the memory kits in the memory settings the GPU is the same and I did swap out this core sage 100 IV to over to the 7900 X system for benchmarking since that does affect thermals of course the GPU driver is the latest from Nvidia 3/8 4.9 for Windows 10 is set to high performance mode for both systems and I'm also running AMD's right and master software on the thread Ripper system and Intel's turbo boost max 3.0 on the skylake X rig let's get into the benchmarks now starting off with Cinebench which is a test suite based on max on animation software cinema 4d it can use all of the CPUs threads to render a seam so the more cores and the more threads the better the score for multi-threaded scores the 1950 X definitely took the lead here with a massive score of two thousand nine hundred and sixty nine just shy of three thousand seventy nine hundred X came in at twenty to ten and then when possible I'm also going to throw in the seventy seven hundred K and 1800 X benchmark results for you so you have them for comparison now single thread mode is also something that you can turn on in Cinebench and it is a good indicator of how cpu might perform while gaming or while running software that isn't multi-threaded like Cinebench is the 1950 exterior shows it's single threaded deficit when it compared to n cell score of 161 compared to the 7900 x with a score of 192 and then finally that 7700 k maintaining a single thread lead with the score of 193 CPU mark is next this is part of the patch mark benchmarking suite dan it runs a series of simulated tasks providing you with an overall score at the end here again we saw the 1950 ex dominate with the score of twenty four thousand four hundred and five followed up by two seventy nine hundred x with the score shy of 23,000 CPU mark also has a single-threaded test that it runs so I dropped that in here for you too and again we can see the same story of when you're looking at these n CPUs of 1950 X and 1800 X you're seeing relatively the same score since these are running it relatively same frequencies around 4000 240 200 megahertz and then they 7900 X and 7700 kwin in single thread mode once again next up is blender which is a free open source 3d creation suite that supports the entirety of the 3d pipeline modeling rigging animation simulation rendering compositing and motion tracking for my test that I ran lower is better here since this is just the time listed in seconds and of the 7900 X actually won this test with a score of twenty five point seven despite not having as many cores or threads as a 1950 X which came in with a score of twenty six point two this just goes to show you that not all software can take advantage of all those threads and when it can't the 7900 X is going to regain a lot of ground on the thread Ripper processor textas pov-ray 3.7 this is a free 3d graphic software it uses the CPU to compute ray tracing to render an image 1950 x dominates here since pov-ray is very good at handling lots of threads and lots of course about a 30% lead over the 7900 x with a total time of 40 4.2 seconds compared to the 7900 x at 57 point 5 we move on to Adobe Premiere Creative Cloud this is the video editing software that I use and I'm rendering a three minute and 54 second 4k video in 3840 by 2160 resolution with a 40 megabit bit rate the 1950s shows again that it is monstrous when it comes to content creation with a total time of 5 minutes and 15 seconds beating the 7900 X by about 30 seconds 3dmark firestrike Ultra is next a test you guys are probably familiar with the overall scores are pretty similar sheer but the physics core is more of the CPU test and here we can see the physics score scaling up with more cores and more threads to the 1950 X as a massive physics score of 26,000 896 and then the GPU test shows again that a little bit better single thread performance can't you get a little bit better performance from your graphics card moving on the 3d mark x spy a DirectX 12 test which I don't think is optimized quite as well for all of those threads and cores because even in the CPU specific test the 79 hundred x manages to pull out a win with the score of eleven thousand six hundred versus the 1950 X's score of ten thousand three hundred and sixty nine now quick note before moving on to the last test Grand Theft Auto 5 I did turn on in game mode here utilizing the rise in master software now in Rise Master you might notice a couple different options at the bottom creator mode and game mode basically this is an option and be enabled to optimize specifically for gaming just a little bit when game mode is turned on two things happen first off legacy compatibility mode is enabled I don't really like that name but it means that one of your two arisin 8-core dies in the CPU is turned off that cuts your number of cores and threads that are available in half but it also eliminates the latency that is introduced in communicating between them now since most games only use at four cores at max you don't really need all 16 of them for gaming anyway and then the second thing is that memory access mode is switched to Numa mode or non-uniform memory access so that each single core works with the memory that's directly attached to it rather than spreading it around and working with all the system memory again that reduces latency for gaming a reset is required to switch between modes but it did allow for performance on the 1900 X that was in line with what the 1800 X experienced but bear in mind that I'm also benchmarking the GTA 5 here at 1920 by 1080 to put more of the load on the CPU rather than the GPU the 7900 X in this case definitely actually had more choppiness if you look you actually see that those 0.1% lows the slowest of all of the frames that were measured the 7900 X actually loses to the 1950 X and that was something that I actually noticed as I was running the benchmark so overall you got a higher average frame rate with the 7900 X but overall I would say the gaming experience was just a little bit smoother with the 1950 X let's move on to power draw and I just have some basic numbers for you here at idle we had a power draw 105 watts for the 1950 x4 55 was what it hit at peak although it was typically drawing a lot less than that depending on what I was doing to do some temperature testing I ran the I 264 stability test stressing the CPU fvu and cache the 1950 X actually performed really well here and I think this was partially due to the frequency it was running out which we'll get to in just a second 58 C was actually the max it hit I was quite impressed especially I'm only using a 240 millimeter radiator considering they shipped us all at three hundred sixty millimeter rad and that thermaltake unit the 7900 ex actually got up to 74 degrees Celsius after about twenty minutes so it was running pretty warm although it was running at a higher frequency as well and I've listed several different frequencies here for the 1950 X as well as the 79 hundred x 1950 X in game mode was running a little bit faster across well not all the cores because not all of them are enabled but I didn't see about 3.5 to 3.6 gigahertz frequency a third Ripper also has xfr extended frequency range just like the mainstream risin parts and it'll actually boost up by 200 megahertz with Red Rivers so 200 megahertz on top of the max boost block of 4000 gets you up to about four point two gigahertz which isn't too bad at all when you're only running at one to four cores and that is why in some of the single trader performance tests as well as I believe in the gaming test we saw performance from Fred thread Ripper which was pretty much what you'd expect from an overclocked 1800 X the 7900 X was also running quite a bit hotter as we've already seen but it was also running faster when on all fours is running at four gigahertz just straight up and it also has that peak turbo boost max mode that it will do on just one or two of the cores and actually saw that hit 4.5 gigahertz max pretty impressive when you're talking about a single core out of 10 at being able to achieve that just with the out of the box settings so just invite all notes to wrap up about thread Ripper in general first off in game mode AMD tells us that it can improve game performance by 5 to 20 percent depending on the game that you're playing and how much latency might affect it there are games that actually suffer from switching to game modes so do take a moment to double check and I'm sure as more reviews come out we'll see more actual tests in different games with dread refer the XS are temperature offset if you guys are call on mainstream risin there's a temperature offset that tricks the system into thinking the CPU is hotter than it actually is that runs the fans at a faster speed lowering the temperature and allowing extended frequency range to extend the frequency range extend extent further make it make it run faster so it's plus 27 degrees for thread Ripper just in case you're interested and if you're using hardware info 64 you can actually see both readout so you can see the actual temperature as well as the offset which is always going to be seven degrees Celsius difference and finally I wanted to point out that AMD told us every single rise in thread Ripper die is abend and only the top 5% of risin dies go into these chips that means of all of the CPUs that they use for mainstream rise and stuff as well as threader fur as well as the epic CPUs as well to get our going into the server stuff the fastest 5% go to thread Ripper that means XFR is going to work better and that means potentially you might see better overclocking over although I think the AMD rise in thread refer 1950 X is an absolute beast especially when you're running software that can make use of all of those cores it cares to workstation tasks and I would definitely give it my recommendation remember though that not all tasks can use all of those cores and threads and in gaming especially we can see that thread Ripper does still fall behind Intel's best even with gaming mode turned on at higher resolutions like 1440 at 4k though the gaming performance would level out between these chips so that's definitely not a deal-breaker especially giving given these products intended uses they're more workstation parts and for gaming although people are still going to use them to game also bear in mind that Intel 7900 X is running at higher sustained clocks leading to higher temps but also making up for some of that performance loss by only having 10 cores rather than 16 I wasn't able to include overclocking in this video but I will be coming back to that to see if running all of thread Ripper scores at 3.8 or 4 gigahertz can make a big difference I'm also going to pick the 12 core thread Ripper 1920 X against Intel 78 20 X so subscribe to my channel if you're interested in those videos hit the thumbs up button on your way out if you enjoyed this one and we'll see you guys in the next one