Media Rendering and Encoding
Cinebench R15
In Cinebench R15, the i9-9900K turns in the single highest single-threaded score we've ever seen. Additionally, the 9900K manages to fall within 4% of the i9-7900X in the multi-thread test, despite its 2-core advantage. That being said, the Threadripper 2950X has a distinct advantage of 34% when all threads are being utilized.
Blender 2.79b
The BMW test of Blender shows a 6% lead for the i9-9900K over the i9-7900X while falling the two CPUs fall within one percent of each other in the Gooseberry workload.
Compared to the i7-8700K, the i9-9900K shows about a 30% performance lead in Blender, tracking almost precisely with its 33% core count advantage.
POV-Ray 3.7.0
The POV-Ray All-CPU test shows a 4% performance advantage to the i9-7900X over the new i9-9900K.
Handbrake
Handbrake encoding is tested with a 4K 100mbps source file, and is transcoded to 1080p with a constant quality of 10mbps in a single pass encode. The encoders used are the X264 and X265 encoders bundled with Handbrake.
While the i9-9900K scores very similarly to the i9-7900X in H264 encoding, the margin between these two CPUs increases to 34% when switching to the more demanding H265 encoding.
X264 Benchmark 5.0.1
The i9-9900K manages to keep up with the Threadripper 2950X in the first pass of the X264 benchmark while losing by 24% in the more multi-threaded second pass.
V-Ray Benchmark
A new test for us at PC Perspective, the V-Ray Benchmark evaluates rendering capability for a given CPU with the V-Ray engine, which is a popular add-on renderer available for many 3D design programs like 3DS Max, Maya, and Revit.
V-Ray's CPU benchmark shows the i9-9900K and the i9-7900X neck-and-neck, while the Threadripper 2950X is around 30% faster.
I want to see how far
I want to see how far i9-9900k can be overclocked until BSOD occurs.
In my particular setup, going
In my particular setup, going from 5.0 to 5.1 GHz all-core immediately bluescreened the system when any load was applied!
This review commits the
This review commits the cardinal sin of chart generation: The colors representing a given product change from chart to chart.
How so? The focus of this
How so? The focus of this review (i9-9900K) is light blue while the rest of the data points are red. The graphs with multiple data points do differ from this the best, but I tried to highlight the 9900K result in blue in every chart.
Take a look at the charts for
Take a look at the charts for Euler and 7z. The color that represents a given product changes from chart to chart.
This threw me off as well. I
This threw me off as well. I ended up opening up this review on 2 screens – one to view legend and one to view charts.
Ah, Sorry! Naturally excel
Ah, Sorry! Naturally excel ruined the charts… I will correct that going forward!
So the Intel i9-9900K is not
So the Intel i9-9900K is not the best CPU for gaming.
It’s undisputably the best
It’s undisputably the best CPU for gaming. It may not the the /optimum/ CPU for gaming if price/perf if your factor to optimise for, but then you’d probably end up with an i3 being the optimum in that situation.
I understand Intel’s not the
I understand Intel’s not the easiest company to like, but this AMD faboyism is idiotic.
One minor correction to the
One minor correction to the chart on the first page: the Ryzen 7 2700X has 16MB of L3 cache rather than the 20MB listed. Unless that row is supposed to be L2 + L3? But this throws the Intel numbers into question…
The PCIe lane count is also a little inconsistent because the TR 2950X is listed as 64, which clearly includes the 4 lanes used to connect to the X399 chipset, while the 2700X is listed as 20, which doesn’t include those lanes bound for the chipset.
And that concludes the nitpicking 🙂
I appreciate these kinds
I appreciate these kinds of nitpicks. The appropriate corrections have been made. Thanks@
Great, that 2700X L3 cache
Great, that 2700X L3 cache figure looks better. I think there’s still an issue with the PCIe lanes though. Here’s what I was attempting to convey:
2700X 2950X
PCIe lanes (inc. lanes to chipset) 24 64
PCIe lanes (exc. lanes to chipset) 20 60
So, my thinking is that 20 was correct for the 2700X as that’s the number of usable lanes after connecting the chipset, but the 2950X should be dropped to 60 as you’ve only really got that many available to peripherals.
And *that* concludes the nitpicking 😛
I’m not in the market, and
I’m not in the market, and while the hardware is impressive I don’t have the scratch for this kind of stuff. My 2600X and RX 480 4GB are meeting my needs at this time. Next upgrade will be the GPU, but not until its interfering with my needs.
TLDR: Cool chip bro….not marketed for me clearly.
I was considering this
I was considering this processor for an upgrade but it seems as if it’s incompatible with windows 8.1, and I refuse to install windows 10 on my main machine.
And 8.1 is better how exactly
And 8.1 is better how exactly ???
It’s not better than 7 but
It’s not better than 7 but 8.1’s EOL is 2023 and has less of that Windows 10 style forcing that always breaks things.
Windows 8.1 can be made to look like windows 7 with third party software to turn off 8.1’s annoying TIFKAM UI nonsense and did I mention that 8.1 has no update forcing that breaks things like Windows 10. And don’t forget the latest windows 10 Update that’s deleting some folk’s user libraries like My Documents, My Pictures, etc. So 7 is EOL in 2020 and 8.1 is EOL in 2023 and 8.1 is really pretty much the same under the hood as windows 7 with some minor tweaking for 8.1 and that crappy TIFKAM UI in 8.1 that can be fixed via third party UI software.
Windows 10 and its Candy Curshing that can not be kept from re-installing with the next forced update along with the Bend Over Here It Comes Again(BOHICA) treatment of forcing on the PC’s owner from the folks in Redmond.
Enjoy your 10 UWP(Universal Windows Peasant) existence!
1. if revit 2019 does not
1. if revit 2019 does not work well on multicore processors. then this is something you should not be using for a benchmark. something is clearly wrong. 2. your running games on these chips knowing that they are coded specifically for intel chips. 3. when has intel made a chip from the ground up? ryzen is a ground up chip, so they are not writing for it yet. they are barely patching games to run on ryzen. this is another bias review for intel.
He could downclock the cpu to
He could downclock the cpu to 2ghz, throw it in a b350 motherboard, and tap dance around in an AMD bodysuit….something tells me you would still claim the review was biased toward Intel
Wow, this is the first review
Wow, this is the first review I read that say 9900k is a good value at 500++ bucks.
Who is this processor for?
Who is this processor for? Genuine question.
Gamers are better off with the previous generation once OCs are applied.
Workstations are much better off with Threadripper, especially since first gen are going for sub €500.
So who is Intel going after with this product that needs a $100 just to keep it from Throttling at stock speeds?
Workstations are better off
Workstations are better off with Epyc and 128PCIe lanes and 8 memory channels per socket. For Folks that need more memory bandwidth a Dual Socket Epyc/SP3 motherboard can be had for around $450-$550(Supermicro H11DSi costs $512) and 2, 16 core/32 thread Epyc 7301s cost around the same price a one Threadripper 2990wx.
So its better to get 2 of the lower cost Dual Socket Epyc 7301 SKUs and have 16 total memory channels across 2 sockets if you are doing any workloads that love more memory bandwidth. The 7301($825.00) is very affordable and the cost of 2 and an Epyc/SP3 Dual Socket MB can come in at a similar cost than TR 2990wx($1,799) and MEG Creator MB($500).
Workstations are better off not with Threadripper and only 64 PCIe lanes and 4 memory channels when one could get a dual socket Epyc/SP3 MB with 16 memory channels across 2 sockets in dual Epyc 7301 configuration. Also with 16 memory channels to populate the user has more options to purchase those very low cost low capacity 4GB ECC RDIMMS($58-$70 per DIMM) and still have 64 GB of total memory or 128GB of memory for 8GB ECC RDIMMs($100-$150 per DIMM).
ECC memory is costly compared to non ECC memory but that’s what workstation users require if they are doing and professional production workstation workloads. Used Server/Workstation memory costs even less but still memory is not cheap these days.
Yeah I’m not seeing how this
Yeah I’m not seeing how this is a gold award.
This should be a silver award
This should be a silver award at best due to the high price.
The i9 9900K is Intel’s
The i9 9900K is Intel’s attempt at trying to maintain its higher markups/margins against AMD’s Zen+/Zeppelin based competition. Intel still has the relatively better IPC and clock speeds against the Ryzen 7 2700X 8 core/16 thread competition. But the relative performance difference that the i9 9900K provides above the 2700X does not justify the added cost. So AMD’s Ryzen 7 2700X will be a better price/performance value even at 1080p with any gaming at 1400p and above an even closer march-up after things become more GPU bound anyways. And really its that latency Advantage that helping Intel in gaming at 1080P.
The i7-9700K at 8 cores/8 threads(Hyperthreading Disabled)
is the better gaming value at 1080p than the i9 9900K and the Ryzen 7 2700X may be a little better at some workloads compared to the i7-9700K but if you are mostly into gaming then really the 9700K($374.00 – $385.00 Intel ARK MSRP) is priced closer to the 2700X.
But still on the Tech Report’s review of the 9900K the Ryzen 7 2700X’s frame times are not really that bad even at the 11.1ms frame times in Hitman where the Ryzen 7 2700x performs a little better than both the 9700K and the 9900K. It’s only at the frame times below 11.1ms that Intel’s 9700K and 9900K really begin to take the lead at 1080P in hitman. So that’s where Intel better latency performance helps and still the 9700k leads there even above the 9900K.
Also on most of the other titles looking at the 11.1ms frame times the Ryzen 7 2700x is not that far behind compared to Intel’s latest until the frame times get to the 8.3ms and 6.94ms frame time ranges where the 9700K and the 9900K trade blows for dominance.
I guess if you are gaming on a high refresh rate monitor at 1080p on ultra settings then Intel may look better but I’ll bet that with some minor settings adjustments that the 2700X would get enough of a higher frame rate that the image quality differences would not be noticable on higher FPS gaming for 144Hz monitors with display adaptive sync activated. It all depends on the game and how that game can be set to target higher average frame rates and most folks are only going to notice high Frame Variance rather than image quality during high frame rate gaming.
Now if only the review sites would do more testng where they lower some settings while trying to get the 2700X to perform like the 9900k in gaming and see if anyone will notice the difference at between 90 and 140 FPS. Because at 1080p that’s not really enough resolution to tell the difference at higher FPS anyways.
Really Tech Report still thinks the i7 8700K was already the world’s best CPU for high refresh rate gaming so that’s maybe still an option also.
Thanks for testing 1920×1080
Thanks for testing 1920×1080
Most reviewers test CPUs at
Most reviewers test CPUs at 1080p these days while using a top-end video card because the CPU will most often be the bottleneck. At 1440p or especially 4K the graphics card is more likely to be the performance limiter and differences in processors won’t show up nearly as well.
Yeah, don’t agree with why
Yeah, don’t agree with why you gave it the Gold Award; undeserved. Silver at best.
“In both 7-zip compression
“In both 7-zip compression and decompression, the i9-9900K only falls to the i9-7900X, managing to best even the 16-core Threadripper CPU.” on page 3…
Are you reading a different chart? The 2950x is much faster in both compression and decompression.
Also.. My 5 year old e5-1680v2@4.4ghz:
Compression: 61,251
Decompression: 65,302
AMD 2950X vs. i9-9900K
I love
AMD 2950X vs. i9-9900K
I love the PC Perspective HW Leaderboard, but while the i9-9900K may deserve to be the Dream System CPU, for multi-threaded desktop workstations it looks to me that the AMD 2950X blows it away(1).
Why not split the Leaderboard into Gaming and Productivity tracks?
(1)
https://cpu.userbenchmark.com/Compare/Intel-Core-i9-9900K-vs-AMD-Ryzen-TR-2950X/4028vsm569025
https://techgage.com/article/intel-core-i9-9900k-workstation-gaming-performance/3/
Worth considering, a few
Worth considering, a few other sites split that way in addition to the ones you mention. For the moment it is the best on the market overall so it sits there but for my next update I will mention TR and X399 as choices for those that work and game.
Great idea, Jeremy!
I know
Great idea, Jeremy!
I know some people game and work. For others their only demanding PC use is gaming. My only demanding PC use is work.
With the current Intel/AMD competition it might be more difficult than ever for one pick to fit all.
Benchmarks aside, the
Benchmarks aside, the difference this chip makes in Revit is noticeable. Coming from a 1700 oc’d to 3.9, the application feels visibly snappier. Enscape, a plugin for real time visualization, also has the model ready much quicker. However, especially with bigger models, say 200MB plus, there is still the occasional waiting to do. Looking at the Task Manager, it seems some more multithreading could speed things up. Hence, if one uses mostly Revit, the 9700K might be just as capable.