Tuesday, May 12th 2020
Intel Core i5-10400 Tested, Significant Multi-Threaded Performance Gain Over i5-9400
Intel's upcoming Core i5-10400 processor, priced at USD $184, with an iGPU-devoid i5-10400F variant priced at $157, could be a serious mid-range price-performance package, building on the popularity of its predecessors, the i5-9400F and the i5-8400. The new chip is 6-core/12-thread, with 12 MB of shared L3 cache, or a similar die configuration to the 8th generation Core i7 series. The chip has the same 2.90 GHz nominal clock as the i5-9400, but increases the max Turbo Boost frequency by 200 MHz to 4.30 GHz.
A PC enthusiast on ChipHell, with access to an i5-10400, tested it on an MSI MAG Z490 Tomahawk motherboard, and compared its performance with the i5-9400F. Among the strictly-synthetic tests are Cinebench R15 and R20, various forms of CPU-Z bench, and SuperPi. The processor posts a tiny 2-5% performance gain in single-threaded tests that scale perfectly with its 4.8% higher max boost frequency (4.30 GHz vs. 4.10 GHz on the i5-9400F). It's the multi-threaded tests where the i5-10400 comes alive, thanks to HyperThreading. It posts massive 35-45% performance gains with CPU-Z bench multi-threaded; a 41.85% gain with Cinebench R20 nT, and 45.05% gain with Cinebench R15 nT. This would bring the i5-10400 within 10-15% of the Ryzen 5 3600X in multi-threaded Cinebench tests.
Sources:
ChipHell Forums, VideoCardz
A PC enthusiast on ChipHell, with access to an i5-10400, tested it on an MSI MAG Z490 Tomahawk motherboard, and compared its performance with the i5-9400F. Among the strictly-synthetic tests are Cinebench R15 and R20, various forms of CPU-Z bench, and SuperPi. The processor posts a tiny 2-5% performance gain in single-threaded tests that scale perfectly with its 4.8% higher max boost frequency (4.30 GHz vs. 4.10 GHz on the i5-9400F). It's the multi-threaded tests where the i5-10400 comes alive, thanks to HyperThreading. It posts massive 35-45% performance gains with CPU-Z bench multi-threaded; a 41.85% gain with Cinebench R20 nT, and 45.05% gain with Cinebench R15 nT. This would bring the i5-10400 within 10-15% of the Ryzen 5 3600X in multi-threaded Cinebench tests.
45 Comments on Intel Core i5-10400 Tested, Significant Multi-Threaded Performance Gain Over i5-9400
It has SMT enabled again which brings 30-35% boost in multithreaded tests (and a fairly minor 200MHz speed bump).
Saved them from being innovative.
The more interesting tidbit is the single core performance. A 7% jump.
Inb4 the obvious, separate announcements of each SKU in 10th gen going forward... and our comments below it ridiculing this. Teeeheee enabled overclocking, you mean like how its done on Ryzen right now? :roll::roll:
The days are soon over...
Can it run cinebench??
A for effort :)
if wasn't for AMD, that 6 cores + 6 hyper threads would be fetch for $100 more.
The 8700 non K is never going to surpass the 8700K. All that has happened since is higher turbo's for lower bases.
Many non K CPUs have lower turbo's. And, Ryzen 3 is stock here. And, this was the day before yesterday. Things haven't really improved much on the Intel side...
The gaming advantage for Intel has become extremely situational by now. As in, only if you chase maximum FPS on a highly single threaded game. Many games are not that anymore; the example below guzzles threads, look at where the 6c6t ends up compared to 6c12t.
With similar core counts, Intel has a very minor advantage, at best, even in gaming. Its time to let that penny drop now.
If we do a more apples to apples with ram OC (this kit is easy to OC to 3733cl15).
3600 180usd
B450 80usd
Crucial 3000cl15 80usd
I5 10400F 160usd
212 evo 30usd
Z490 120usd for cheapest budget
Crucial 3000cl15 80usd
You will get slightly better perf with Intel for 50usd more.
You could get a B460 and use stock cooler and beat AMD pricewise, but performance in most games will be lower due yo 2666 ram.
As for the graph, from my understanding, you need 6 threads for that game. More is fine but doesn't seem to help, and it looks like the higher frequencies are accounting for the better performance numbers towards the top of the graph, which oddly doesn't include the 9900K/F/S or even 9700K.
Number 2, I never said an 8700 would beat an 8700K, I said a (recent) stock intel cpu is comparable to an overclocked RYZEN cpu (for gaming). Please reread that part. Um, tuning ram is considered overclocking. Yes, going outside of specifications on a ryzen PLATFORM is considered overclocking. Also, 3600 MSRP is 199$, while 10400F MSRP is 157$. As for the motherboard, huh, i guess the motherboards arent even out yet. Also, why do you need to purchase an extra cooler? I doubt the 10400F will throttle on the stock cooler (though it isn't a bad idea to get a cooler generally speaking). Like i said, a majority of people don't want to tune every last timing on their system, intel OR amd, so OVERCLOCKING/TUNING ASIDE, Intel almost always has better performance in games than Ryzen, at least in a similar price range. Think of it this way, what percentage of the population of people who own a car buy it so that they can tune it to be better than another car? Not a high percentage. Likewise, a relatively low percentage of system builders buy a system to tune it to be better than another system.
Here's the dots I connect:
- Ryzen CPUs you don't typically OC anymore, they're best on their own.. and yeah. RAM overclocking... so XMP is also an overclock now then :p Definition needs an update maybe.
- Intel CPUs non K cannot be overclocked, so you're talking stock vs stock anyway
- Intel K CPUs are extremely costly because to extract higher than Ryzen performance, you need to have sufficient cooling, cost can easily ramp up to 150% of a similar tier Zen CPU for the package. High power draw also affects case temps, and we're talking about significant TDP gaps here.
- Intel non K CPUs sacrifice base clocks for higher turbo's these days, they may burst a bit faster (nanoseconds of profit...) but under sustained loads, Ryzen will be faster every time.
- Ryzen SMT seems to scale better across the board
So what you have with a non-K Intel CPU, even if price is similar, is a less well rounded CPU and what you get in return is a highly situational advantage in a steadily decreasing percentage of game titles. The tables have quite definitively turned into AMDs favor, if you ask me. Comparable, yes, but certainly no longer the optimal choice. We have not even touched yet on Intel's security problems and bandaids which still, until recently include microcode updates through Windows Update, or the fact you cán indeed undervolt or OC your Ryzen CPU where the non K is unable to; or that motherboards are not dead ends.
Tuning a car is a terrible comparison as it voids warranty on most cases and requires a lot more work than using dram calc. You can OC ram and not voud your warranty unless going beyond recommended voltages.