Gadgetory

welcome back to harbor unboxed this is my second time around filming this introduction once I finished editing and putting this whole video together I was quite shocked at the mammoth that it turned out to be so if you're after a quick five minute video or something that just summarizes things really quickly this video isn't going to be for you we sort of cover everything or everything that was humanly possible in the last week I did multiple not even 16 hour days thing there were 18 hour days but anyway it doesn't matter it's a big long video so you've been warned I'll add a video index somewhere on the screen I'm not sure where we haven't done that before but I'll do that and also pin one down in the comment section below so you can jump to bits go over things again if you really want to and yeah but anyway you've been warned let's get on with it for those of you who missed the unboxing we have the thread Ripper 2990 WX I'm going to say that a hell of a lot in this video so also be warned about that and the 2950 x also be saying that a lot but not quite as much so I've got those two on hand for testing and these two CPUs are very different and they do share the same DNA but they are quite different we'll go over that in a moment they also target different segments of the market and really probably deserve their own reviews but for the sake of time the time we had available I've just lumped them together into this one big fat review but they'll no doubt get their own individual analysis over the coming weeks so as we learn new things we'll put that into sort of maybe a two separate mini reviews a week or two in the future the second gen interpreter CPUs were announced earlier this year during Computex and since then everyone's attention has been focused firmly on the 32 or 64 thread part now knows the 2990 WX coming in at $1,800 us there will be two models in WX series and for those wondering the W signifies that this is a work station series and the X while the usual extreme nonsense I suppose along with the 2990 WX there will also be a 24 core 48 thread model known as the 2970 WX though that model won't be available till October but we do have some simulator results in this video that we can go over although the 2990 WX has been receive the attention the 2950 X is the real hero of the lineup and basically what we have here is a refined 1950 X at a $100 US lower launch price as was the case with the second gen rise in five M rise in seven models these new thread ramparts feature reduced cache and DRAM latency with support for slightly faster memory so they're based on these n+ architecture which uses the 12 PL protests from Global Foundries the 2950 X features the same layout as the 1950 X and this means it comprises of two active Zeppelin dies each packing out calls to memory channels and 32 PCIe gen3 lanes when using ddr4 30 20 memory the Infinity fabric throughput between these dies is roughly 50 gigabits per second as was the case with the 1950 X the 29 50 X can be configured in one of two ways using uma uniform memory access which aimed he refers to as distributed mode in their rise and master software under this mode the processor acts as a single unit and this means threads and DRAM transactions are distributed evenly across the entire chip to maximize bandwidth but in turn this does increase latency which isn't ideal for things such as gaming therefore it is possible to enable Numa or non-uniform memory access which aimed II refers to as local mode in the rise and Master software they call this a local operating mode as the processor is separated into two domains and attempts to pair active cores with local DRAM rather than accessing memory via controller in a separate die which comes with a rather hefty latency penalty the 29:19 WX on the other hand is a very different beast it consists of not to Zeppelin dies but rather for enabling up to 32 cause however on the X 399 platform aimed II has imposed some limitations to avoid cannibalizing their single socket epoch CPUs the biggest of these limitations being that there are still just four memory controllers although there are two more Zeppelin dies AMD calls these compute dies basically what that means is they have no local PCIe or DRAM access for that they must travel through the Infinity fabric to the i/o dives so they would be the Zeppelin dies with direct access to PCIe and ddr4 memory as there are twice as many dies in this configuration the inferior fabric is also hard so the throughput between dies is just 25 gigabits per second assuming you are using ddr4 3200 memory because of this design which sees 2 dies without direct access to DRAM it means that unlike the 2150 X the 2990 WX uses numerous cluesive Lea AMD says this quad Numa configuration has allowed them to create the world's first 32 core consumer processor and just as importantly it has allowed them to do it while maintaining backwards compatibility with existing tr4 products there is however a fairly obvious drawback that's had me a bit concerned since we first heard about this 32 core model we always knew that the first gen third Ripper CPUs had the potential to offer up to 32 cores so this isn't some kind of radical breakthrough for AMD with their 2nd gen series the original thread ripper chips did in fact have 4 dies though two of them referred to as dummy dies as claimed by AMD but we always knew they were just defective Zeppelin dies that were disabled but physically there as after all thread Ripper is just an epic CPU repurpose for the high-end desktop platform that's it I don't mean to sound like I'm downplaying anything here epic CPUs on the desktop is very much epic anyway for this 2nd gen thread Ripper series amy has enabled those extra dives to create the 24 core and 32 core models the problem however is one of memory bandwidth there simply isn't going to be enough of it as I just discussed we still only have quad-channel memory access so the memory bandwidth remains the same but now we have twice as many cause to feed this is likely going to make an already very niche product even more focused so keep that in mind for testing I've got a truckload of data and while I tried to include as many CPUs as possible I did run out of time to go back and retest the core i7 87 or ok this is something I can do in the follow-up content that'll come over the next week or two the good news is I have results for a number of intel's high-end desktop processors such as the core i 979 ATX II and the 79 60 X for example rather than discuss the system specs and go over all that stuff individually I'll just throw them up on the screen and some tables and if you want to examine them a bit more closely you can hit the pause button basically all systems were configured with 32 gigabytes of ddr4 memory most of them are running at the 3200 spec using the XMP timings so there's been no tuned memory timings in any of these tests so having said all that let's get some results ok so we might as well get this one out of the way first Cinebench r15 as many of you are probably aware now given that AMD did leaked the results the 2990 WX achieves a score of just over 5,000 points in its stock out of the box configuration that makes it a whopping of 52 percent faster than the core I 979 80 XE and at this point you're probably wondering what the hell I was on about when I said there were some drawbacks to this design but hold that thought we will get to that later on anyway in this rendering benchmark the 2990 WX has no trouble blowing socks completely off the 2950 X is no slouch either though it does only improve upon the 1950 X by a mere 5% margin next up we have another rendering benchmark though this one is based on real-world software the Corona renderer has been used to test workstations with over 64 cores so it scales very well and has no problem using all 32 cores and 64 threads of the 29 and 90 WX here we do again see some pretty breathtaking rendering performance from the 2990 WX as it took just 41 seconds allowing it to completely test 40% faster in the 29 50 X not perfect scaling but it is still an impressive result this also meant that it was 28% faster than intel's current flagship core I know in part so yeah that's also very impressive also this time the 2950 X was just 4% faster than the 1950 X so another small game there but again all the same moving on I fired up the rise in graphic workload in blender now this is a relatively quick test for these high-end CPUs and we see that the 2990 WX took just 8.3 seconds this meant it completed the workload 36% faster than the 2950 X and 31% faster the kora 979 ATX e again an impressive completion time for the 2990 WX but it has to be said for a doubling of cause we are only seeing a 55% boost in overall performance when compared to the 2150 X and there is only a minor clock speed difference between the two so 55% more performance for a 100% increase in cause isn't that impressive that said I am keen to see how the two compare in a workload that takes significantly longer than a few seconds for the 2990 WX to complete so let's check that out disappointingly though we do see this much more complex gooseberry workload was less favorable to the 2990 WX to be fair it did still uproot the 79 80 XE and kick it spins in but it was only able to reduce the completion time by 28% when compared to the 29 50 X it was able to reduce the render time by 20% we compared to the more expensive 79 80 XE so that's still obviously quite a great result for AMD still it is a troubling sign that in what should be an optimal workload for the 2990 WX we're only seeing a 38% increase in performance for a 100% increase in cause okay so poverty is the last rendering benchmark we're going to look at and this one does bode well for the 2990 WX here we are able to reduce the rendering time by 40% we compared to the 2950 X and this meant it was 65% faster so again not amazing scaling but at 65 percent it is much better than what we saw in Corona and blender it was also 57% faster than the core o 9 79 80 XE so a massive win there for AMD I've included the heavy multitasking results from real bench which runs image editing video compression and rendering tasks simultaneously AMD's new 32 core protest saw a peak load of 70 percent but for at least half the test the load was down around 20 percent so that is worth noting here we see the 2990 WX providing a surprisingly poor result taking 43 seconds to complete the workload this made the 32 core processor lower than even the 1950 X and this bench market really was the 2950 exit impressed matching the core Oh 979 60 X and 79 80 XC this meant the 2950 X was able to complete the heavy multitasking test six percent faster than 1950 X so we get a pretty great result there PC mark 10 has been included just to check general performance there are no oddities to speak of here the 2990 WX is comparable to the core I know in 79 60 X which is perfectly acceptable for a super core heavy CPU the 2950 X does do much better though as it comes in just behind the 1920 X okay so moving on to some spreadsheets excel is a real-world application that we have been using a benchmark CPUs for years and years now so I thought why not try the Monte Carlo simulation with the second gen thread Ripper CPUs as expected the 2950 X is slightly faster than the 1950s shaving 2% off the completion time meanwhile the 2990 W X is slower than the 2950 X are basically matching the previous generation 16 core parts so a bit of a disappointing result there AMD's Rison architecture has always excelled for decompression work and we see that the 2990 WX is no exception absolutely smashing at the 7-zip 32 megabyte dictionary decompression test with a throughput of 187,000 mips that said it was just 31 percent faster than the 2950 x despite packing 100% more cos still AMD's SMT implementation works wonders in this test and as a result the 2990 WX was 116 percent faster than the kora 979 80 XE but as impressive as the decompression performance is compression performance leaves a lot to be desired for compression work the core I 970 960 X was 9% fast in the 2150 X which isn't a bad result for AMD there and it was nice to see the 2150 X providing an 8% boost over the 1950 X the 2990 WX though well that one fell into a complete heap despite 100% core utilization throughout the test performance was less than that of the 1920 X and just 23% higher than the 8-core 2700 X this memory sensitive test gives us our first look at how Pauli things can go for the 2990 WX it's pretty ugly stuff and short of shutting cos completely off there appears to be no fix I'll look at this issue a bit more closely later in the video but for now let's continue with the application benchmarks veracrypt provides a built-in benchmark that allows users to test out how various encryption algorithms perform on a given system these tests aren't run from slower local storage but rather system memory and this proved to be a problem for the 2990 WX please note this benchmark uses 100% of the 2990 WX all 64 threads are fully loaded so the odd performance isn't a result of underutilization there even if that was the case it should at least match the 29 50 X the problem is what we have here is another memory intensive benchmark though I should note that it's not that memory intensive the 50 megabyte test only loads 50 megabytes of data into the system memory so this has little impact on the 2990 WX even though it was slower than the 29 50 X and even the 1950 X the one gigabyte buffer test though loads one gigabyte worth of data into the system memory and this proves to be a massive issue for the 2990 WX it's 32 cause become choke to the point where they're significantly slower than these 16 cause of the 29 50 X offering roughly half the performance and so that's pretty shocking to see we see a similar story when encoding with handbrake although the 2990 WX is significantly only utilized in this test maxing out at about 30% it still doesn't make up for the poor performance under these conditions you would expect it to at least match the 29 50 X and it simply fails to do so here the 29 50 X was 15% faster than the 32 core processor though I should note that it also offered no real our performance gain over the older 1950 X meanwhile the 32 core processor was only able to match the 1920 X which is obviously a massive disappointment speaking of massive disappointments this one was particularly disappointing for me we found when encoding with Adobe Premiere that the 2990 WX took 35 percent longer to complete the workload when compared to the 16 core 2950 X the good news here though is the 2950 X was 7% quicker than the 1950 X and that made it just 7% slower than the core I 979 60 X so pretty great result therefore the 2950 X and well I don't know how else to put it a disastrous result for the 2990 WX although the export test only utilized the 2950 X to the tune of about 90% which to be fair is quite high the 2990 WX was only seen to be utilized at around 60% which you know isn't particularly great for the warp stabiliser test though we ran a dozen instances simultaneously and that was able to actually max out all 32 cores and 64 threads of the 2990 WX despite that the performance was still mighty underwhelming and here the 32 core process I was against slower than last year's 12 core model and I only managed to match the Intel 10 core sky like X part so pretty shocking results therefore the new 32 core processor but the good news is the 2950 X wasn't nearly as disappointing it was able to shave five percent off the 1950 X's completion time to take out top spots so a pretty amazing result therefore the 16 core processor okay so why is the 2990 WX so disappointing in a lot of these tests well as I alluded to earlier it is 100% down to memory bandwidth much more so than core to core latency or memory latency here we can see the sustained memory bandwidth phage processor you'll notice that the 2990 WX is a little down on the 2950 x and that's due to the added latency that dies without a memory controller incur it's a 7 percent drop in bandwidth but that alone doesn't explain the performance issues seen so far just to confirm those results I did also test with a 264 and here is the memory copy performance again the 2990 WX was down 7% on the 2950 x but that alone doesn't explain the misery performance in the encoding compression and encryption benchmarks for that we need to look at memory bandwidth per core not the entire processor but rather individual cause arranging these results by a single threat we see that with just one core active the Rison cpus enjoy a tremendous bandwidth now please note the performance of each core within the CPU is measured individually and the result you see here is the average bandwidth across all individual cores so the 2,700 X and 2950 X both second-gen parts deliver the same 29 gigabytes per second then the first generalizing parts deliver between 24 and 25 gigabytes per second and then we have the 2990 WX at 20 gigabytes per second this is why we saw a slight drop in total memory bandwidth in the previous test the margin is amplified here showing the 2990 WX to be almost 30% slower as we're not limited by the ddr4 memory in this instance the reason the single core bandwidth is down is due to the fact that 16 of the 32 cores aren't connected directly to the memory and therefore do suffer increased latency finally we said that almost all the skylake experts our limit is just 14 gigabytes per second though this is less of an issue as fourteen gigabytes per second per core is essentially overkill and here's why if we rearrange this graph by the all threads active result the arrangement changes quite a bit now for these results all CPU cores are actively accessing system memory and we're showing the average throughput of an individual core essentially with the CPU running at full steam in a memory intensive workload this is the typical amount of bandwidth each core has at its disposal this here is the very problem the 2950 ex enjoys a bandwidth a four point four gigabytes per second per core when maxed out and this is why the 14 gigabytes per second we saw with just a single core active on the Intel CPUs isn't an issue since the maximum sustained bandwidth of a scale like X processor is around 64 gigabytes per second so technically with just 5 cause active in an extremely memory intensive workload you are going to use up all that bandwidth and once you start adding more cores you start to see a drop in efficiency as they fed enough data naturally the more cores you have the worse you're going to end up being in this test without increasing the overall memory bandwidth with octave channel memory the 2990 WX would indeed better match the four point four gigabytes per second per core of the 2950 X but with just quad-channel memory that figure is halved well a little over half due to the increased latency so bit of a double whammy there in the end just shy of two gigabytes per second of bandwidth per core just isn't enough and we see the problem this causes when running memory sensitive applications such as Varrick for example okay so before we move on to overclocking power consumption a few other tests let's quickly go over gaming performance right so AMD has stressed heavily stressed in fact that the 2990 WX isn't a gaming CPU though they have gone around now and put on their website that it is a gaming CPU but we'll touch on that later in the video anyway it's fair to say that the 2990 WX really isn't a gaming CPU but I know you guys are going to want to see how these CPUs performing games and while the 2950 X does sort of make sense for gaming having said that though games are still struggling to utilize the eight core 2,700 X so they are going to benefit from 32 cores and they certainly aren't going to benefit from a design that the 29 and 90 WX features even so the 2990 WX isn't terrible ashes of the singularity and whilst slower than the 2,700 X the performance is perfectly fine we also see that the 2950 X takes a small step from the 1950 X tacking on a few more frames and overall delivering a great gaming experience in this title also remember that we are using a gtx 1080 TI at te p with the higher quality preset which is two steps down from the maximum quality had we used the extreme or crazy presets than at least the top half of the graph would have been heavily GPU bound ok so these results look a lot more brutal for the 2990 WX though I should note that gameplay was still very smooth with no stuttering and frame rates did remain well over 60 fps so there is that still the 32 core model was well down on the new 16 core 2950 X which averaged an impressive 153 FPS then moving on we finest a story when testing with f1 2017 the 2990 WX looks pretty much like a low-end Pentium processor but at least it was smooth and playable then we have the 2150 X which did manage to match the 2700 X and this meant it wasn't a great deal slower than the most expensive sculling xcp use the thread Ripper CPUs do play very well in Assassin's Creed origins taking out the top three spots grant it at most they were just a few frames fast and the Scarlet X parts but still it is a great result as expected though the 2990 WX does tank here and it is noticeably slower and the other process is tested before we move on from gaming here's a quick look at the half legacy mode for the 2990 WX this is a feature that can be accessed in the Rison master software essentially it's a down core function that disables half of the dies in the 2990 WX so two of the dies and therefore 16 cause there is also a quarter mode which basically turns it into a 2700 X these legacy modes are for software that doesn't work well with all cores active software like games for example as you can see in the half legacy mode the 2990 WX sees a 10% boost in performance when testing with ashes of the singularity still a sure the singularity wasn't really a bad result what was a bad result was what was seen in f1 2017 here we see a massive 135 percent increase in framerate and this allows the 2990 WX to act very much like the 29 50 X so the gaming performance can be fixed by turning the 2990 WX no 2150 X but that's hardly a practical solution I should make it very clear that in order to enable a legacy mode you need to execute a full system reset so it's really only something you'd use out of absolute desperation and it would only be a very temporary solution as I assume anyone buying a 32 core processor once 32 cores and not 16 all right if you're wondering what the half legacy mode does for applications here's a quick look at that firstly memory bandwidth is increased massively and it's actually now faster than 2950 X and this is because we have two dives using Numa opposed to the 2150 X which has two days using uma though that's just the default so you can change the 2950 X to use Numa and that would see the same seventy five point nine gigabytes per second just to confirm that increase in bandwidth I also around a to sixty four and here you can see the peak throughput now exceeds eighty eight gigabytes per second on that note please note that this figure is higher than the seventy six gigabytes per second reported by SCI software because here we're reporting the peak throughput rather than the sustained throughput here we see that the inherent issues with the twenty nine ninety WX is solved and that issue is of course per core bandwidth with the more memory bandwidth overall and half as many cores to feed we see a little over twice their bandwidth now available to each core now although the bandwidth per core has improved we do only have half as many new cause and this means in workloads where the 2990 WX did do well previously it's going to be a lot slower and now produce similar performance to that of the 2950 X is that essentially what it is now we do see that and blend up the same is also true for povery basically the 2990 WX is delivering 2950 X light performance not that surprising however where the 2990 WX was slow in the 2950 x we are now seeing comparable performance as shown here when retesting with handbrake another example is Adobe Premiere Pro CC the warp stabiliser workload and here the 2990 WX in the half legacy mode doesn't quite match the 2150 X but the results are certainly much closer now so that's how the half legacy mode works that's not practical or particularly useful in my opinion but the options there if you get desperate enough moving on we have some power consumption figures and here we see these slightly higher clock 29 50x consumed 10% more power than that of the 1950 X under full load in the corona benchmark and this placed it on par with the core i9 is 70 900 X then we have the 32 cored 2990 WX pushing total system consumption up to 383 watts which is 19% more than that of the core I nine 79 80 XE still that is only a 36% increase over the 29 50x at least for the total system load so that's not bad given that you do have 100 percent more cores which do to complete this workload 40% faster before moving on a quick note on temperatures using the Wraiths Ripper the 29 90 X around and a very cool 59 degrees though the fan did spin up to 2,300 rpm and while not overly loud it was clearly audible the 2950 X ran a few degrees cooler at 56 degrees and this allowed the 120-millimeter found in the Wraiths Ripper to spin around 200 rpm low which did help reduce the operating volume okay so time to overclock and please note I didn't have a huge amount of time to delve into overclocking and it certainly isn't the focus of this content we can certainly follow up with a more in-depth look into overclocking and future content for this review I tried two methods of overclocking one it using the rog Zenith extreme using the Isuzu precision boost override and then another one using the MSI Meg external creation using the old fashioned multiplier and fixed voltage method both methods resulted in fairly similar multi-core performance though it has to be said the asou speeds significantly better single core performance here we see the fixed frequency overclocked yield slightly better results in the blender test though I should emphasize only very slightly the 2950 X was able to complete the Orenda 12 percent quicker while the 2990 WX was just 4 percent faster which which is disappointing given that we saw a 20% boost in Cinebench ok so these are some pretty alarming results overclocked to the 2990 WX saw at least an 81 percent increase in total system draw and up to 97 percent with the fixed voltage here we can see why fixing 32 cores at 4 gigahertz using one point 4 volts is a bad idea the idle consumption was 192 Watts more than that of a stock rising 7 1800 X at full steam and about what the 1920 X draws under load using the asou speed the idle draw is roughly half down to 95 watts but even say the low consumption still hits an isle watering 694 watts in this test I say in this test because I observed total power draw hitting 780 watts in Cinebench and 884 watts for the fixed voltage overclock this is probable AMD suggests that if you plan on overclocking the 32 core processor you pack at least a 1000 watt power supply interestingly despite promoting overclocking on these unlocked parts aim D says that if you overclock or use P Bo you will void your warranty not sure how they'll enforce that one but they are claiming that any form of overclocking will void your warranty on these second gen fedra per CPU so that's something to keep in mind just quickly here are a few simulated benchmarks using the down core function to turn the 2990 WX in no.29 70 WX the 24 core part and then the 29 50 X into the 29 20 X which is of course the 12 core part 429 70 WX should score around 4,300 points while the 12 core model will be good for around 2600 points in the corona benchmark 429 70 WX sits between the twenty nine ninety WX and Cora nine seventy nine eighty XE while the twenty nine twenty X is a fraction faster than the 1920 X as expected then when testing with blender we see at the twenty nine seventy WX took just five percent longer to complete this test which isn't bad given it has 25% less cause this might indicate the 24 model won't suffer nearly as badly as the 32 core model when all cores are fully utilized and accessing system memory but I'll have to look into that one a bit more then of course the 29 20 X will be a slight upgrade from the 1920 X as a bonus round I've included a he per memory scaling data for the twenty nine ninety WX ranging from ddr4 1866 up to the maximum spec I could get to work which was ddr4 3400 here we see consistent scaling right up to 3400 and how I've been able to get faster memory to work it's likely we would have seen a steady increase right to 4,000 and beyond this right here is why Cinebench r15 is a completely useless application for measuring any kind of memory performance and it's also why either 2990 WX smashes it out of the park in this test it's just not at all memory sensitive yes you can adjust timings and see small gains but yeah you're not gonna see great gains Krone is another benchmark that just really isn't that memory sensitive either at least again within reason it's not until we drop down a ddr4 1866 which isn't really a thing I suppose that we see a major drop-off in performance from ddr4 29:33 and Beyond there's really no gains to be had 7-zip is a memory sensitive application and this is best demonstrated by the decompression test however it was the compression test that really hurt the 2990 wx-- and therefore don't really see the results that we'd probably expect to see here perhaps the increased latency is also an issue for this test I'm not 100% sure on that one we will need to do a bit more digging to find out what's going on well that was plenty of data though in some ways I feel like we're just starting to scratch the surface on this one especially with the 2990 WX over the next few weeks so do want to spend as much time as I can with this processor looking into various productivity workloads um see what I can throw at it and see if I can find anything that's not a rendering workload that performs really well anyway for now let's have a look at a couple of price versus performance scatter plots before I wrapping this long review up let's take a look at the poverty as the 2990 WX had it the best showing here it's much faster than a 70 night ATX eat while costing slightly less so that's obviously a good thing and that means for rendering tasks the 2990 WX really is the bee's knees and while it might not offer the best value overall it certainly provides the best value at the top end of the scale the 2950 x destroys the kora 970 900 X offering significantly more performance at a reduced price really the only better option here is the heavily discounted 1950 X still in poverty second gen 3 Ripper CPUs look very good the only other price versus performance scatterplot that we're going to look at is premiere and this pretty much sums up the 2990 WX when handling non rendering tasks here we see how badly things can go for AMD's new 32 core processor when encoding with premier the 2990 WX is considerably worse value than the core I know in 79 80 XE and therefore rather CPU tested you're basically looking at thread ripper 1920 x performance for almost four times the price thankfully though the 2950 x is again a resale adoption and office pretty much the best bang for your buck of any of the new CPUs tested well the twenty nine ninety WX certainly an unusual beast in short aim DS made a thirty-two core CPU that appears very good for rendering and that was about it in our standard test suite the memory limitations do seem to make the twenty nine ninety WX a a very focused product and I'm not sure how many users are going to find it useful for powering their workstations given the inconsistent results that said I am in the process of conducting some extensive multitasking benchmarks and while I am still seeing some mixed results I think overall what I am seeing is painting a more positive picture I would also just like to note that Amie themselves have admitted that this is a focused product and that's why they changed the name to include the W before the X so it's not just the 2990 X and they also show in their review guide that the 2990 WX beats the 79-80 XE in applications such as or benchmarks such as Corona a blender Cinebench pas vrai mayor and a dope dimension and of course what are all those programs have in common their rendering applications of course meanwhile the 2950 X was tested against the Coraline 7900 X in programs such as handbrake TrueCrypt 7-zip premiere and then of course a few other rendering applications so all the 2950 X can clearly do it all as we've shown in this video aim DS well aware that the 2990 WX can't I think that's probably the best way to put it the only issue I have is that they're not exactly telling you that at least not by the information that I found on their website and rather than give you a chance to work out that the 32 core model might not be all it's cracked up to be for your particular workload they took pre-orders for only that part a week in advance I did voice these concerns with my AMD contacts over the weekend and they assured me that anyone who is unhappy with how the 2990 WX performs for their particular workload can cancel their pre-order no money should have exchanged hands at this point but if it has they can refund and get something else maybe the 2950 ex so we're not ready to close the book on the 2990 WF just yet but given that there are some obvious shortcomings we feel the $1,800 us asking prices it's just a bit much the core I know in seminar ad XE is already grossly overpriced as it is but at least it does do everything really well and doesn't really have any weaknesses other than the price itself which is obviously quite a big weakness in terms of price versus performance which is why we much prefer the 2950 X but it doesn't suffer any performance issues where it's much slower than the lower core parts in its product family of course the 2990 WX has all the usual problems a super core heavy desktop processor would happen that is utilization finding software that can fully utilize it but of course the key key here is the software can't be really heavy on memory access or it will cripple the 2990 W X to the point where it's a slow as a mainstream desktop CPU this ultimately is the issue with the 2990 W X where perfectly fine with most software such as premiere for example not taking advantage of 64 threads but in that case the 2990 WX should act like a 29 50 X and as we've seen that's not always the case and sometimes it is much slower so how do you know if a particular workload will cripple the 2990 WX or run like a dream unless you have someone test it out for you test drive the CPU you really don't know and it's not enough for them just to test the software you'll be using it's that sensitive that you really need the kinds of workloads tested that you're using I think AMD should have just done a better job of stating what the 2990 W X is designed for rendering workloads show where is really good and I'm talking about their website here not what they've told reviewers or anything like that forget the media I'm talking about telling the consumers for the pre-orders that it does have some shortcomings in heavily bandwidth dependent workloads the performance may be lower than expected see I think stuff like that would have just helped clear this up based on our testing we've found it to pretty much be the thread Ripper Pixar Edition or maybe instead of w they could have called it the 2990 CB edition that would have made more sense anyway like I said we're just not ready to call it yeah the 2990 WX still has more to prove but at this point I think it is quite clear that I personally won't be upgrading my workstation to the 32 core processor which is a bit of a disappointment still I will be getting an upgrade and that upgrade will be the real hero of the second gen third Ripper series it's no secret that I really liked the 1950 X and the 2950 X is simply a more refined version typically offering five to eight percent more performance while coming in a hundred dollars u.s. less at launch so great by that and right now though the 1950 X that is coming in at 780 dollars u.s. so that one is a cracking good buy right now but still considering the 2950 X's introductory price it's it's really amazing value I have nothing bad to say about the 2950 exit does everything really well and in my opinion it takes over from the 1950 X as the ultimate high-end desktop CPU I had really hoped that my editing rig would be receiving the 32 core part but as I said before those plans have now been scrapped and I will be moving to the 29 50 X instead of course if you have a 1950 X already like I did then the 29 50 X doesn't really offer a noteworthy or worthwhile upgrade as I said it's only about five to eight percent so it's not really worth scrapping at the 1950 X yet so yeah your CPU hasn't been rendered obsolete just yet and that is going to do it for this one I hope you guys did enjoy this video I know it was a very long-winded look at the second generation thread Ripper CPUs but I think it was worth it and if if you think so as well please feel free to the like button put a tremendous amount of time into this one and subscribe if you want to see more content like this and if you appreciate the work with your hair unbox then consider supporting us on patreon thanks for watching I am your host Steve and I will see you again next time