Ryzen 5 2600 vs. Core i5-8400, 36 Game Benchmark Battle

[IEC]

I don't see anything wrong with what Thala said? Its pretty much spot on. Stock vs stock, the 8400/2666 is a bit faster than the 2600/2933, at a slightly lower overall platform cost. They are both decent at gaming but if you had to pick a 'winner' then it would have to be the 8400, you can't go wrong with faster and cheaper, right?

The 2600 does indeed have the potential to overtake the 8400, but with an additional outlay of approximately $80 for an aftermarket HSF and B Die memory capable of CL14 3400 as shown in the HWUB video

Mark's point is the AM4 platform has superior hardware support and upgradability. Even the cheaper B350 (and soon B450) boards have no limitations beyond VRM for overclocking CPU/memory. All boards support at least 3200 speeds and even my $60 MSI B350 microATX board supports 3600 CL16. In addition to supporting new Zen processors until at least 2020 with UEFI updates. And virtualization. Very little is "tiered" away on the AM4 platform, unlike the Intel equivalents. Z370/B360 are dead ends, and that i5 still has 6 fewer threads.

For people who don't upgrade frequently, it'd make more sense to go for the 6c/12t option. And spring for an extra $20 on a 2600X for the upgraded stock cooler and significantly higher boost speeds. With sales as low as $140 for 3200 CL16 lately, it's possible to get 3000 MT/s or faster memory on a budget. Which would already be beating the 8400 in gaming at similar cost... and have an extra 6 threads to boot.

 

[frozentundra123456]

Yea, the article itself said it would cost about 80.00 to overclock the 2600 to 4.2 ghz and use fast ram. I dont even think we have enough information yet to know how common a 4.2 ghz overclock is on the 2600. The absolute highest guaranteed overclock of the 2600x (maybe better binned than the 2600) on silicon lottery is 4.25, so you know 4.2 is right at the edge. Silicon lottery does not say what percentage of the 2600x can reach 4.2 or 4.25. Even assuming that the 2600 can reach 4.2, the extra cost would allow an intel user to move up to the 8600k with change left over. A better choice for ryzen would be to simply buy a 2600x and not bother with overclocking.

 

[frozentundra123456]

The 2600x is a bit faster than the 8400, for $20 more than the 2600. With the stock heatsink. The 8400 is also on a dead end platform. The platform cost on the 2600x is currently a bit higher. Should be closer to even when the B450 boards appear. Or pretty much even using a B350. Not saying the 8400 is bad, it isn't. But too many roadblocks, like no overclocking at all. Not even on the ram unless you go for a more expensive 370 motherboard.

That being said, I doubt anyone could tell the difference actually USING computers built with either of them, unless you are rendering, modeling or other tasks like that.

How do you know the 2600x would be faster than the 8400? No data is presented in the article under discussion to prove that. The 2600 at all core 4.2 is only slightly (3.4%) faster (with fast, low latency ram as well), and the 2600x will certainly not run 4.2 all core.

 

[Markfw]

How do you know the 2600x would be faster than the 8400? No data is presented in the article under discussion to prove that. The 2600 at all core 4.2 is only slightly (3.4%) faster (with fast, low latency ram as well), and the 2600x will certainly not run 4.2 all core.

Wrong....This part "and the 2600x will certainly not run 4.2 all core"
If the 2600 will do it, why not the 2600x ? This makes no sense.

 

[frozentundra123456]

Wrong....This part "and the 2600x will certainly not run 4.2 all core"
If the 2600 will do it, why not the 2600x ? This makes no sense.

I was talking about all core turbo on the stock 2600x, in reference to scannal's post that the 2600x is faster (I assumed he was talking about stock) than the 8400. If you can show me documentation that the 2600x has a 4.2 ghz all core turbo, I would be most interested to see it. And now that you brought it up, I am still have reservations that the majority of Ryzen can even reach 4.2 overclocked. Obviously, as shown by this article and silicon lottery, some can, but I have seen no information yet as to what percentage can reach this frequency. Obviously, since the absolute top clockspeed on silicon lottery is 4.25, that range is at the ragged edge for even the best of chips.

 

[IEC]

I was talking about all core turbo on the stock 2600x, in reference to scannal's post that the 2600x is faster (I assumed he was talking about stock) than the 8400. If you can show me documentation that the 2600x has a 4.2 ghz all core turbo, I would be most interested to see it. And now that you brought it up, I am still have reservations that the majority of Ryzen can even reach 4.2 overclocked. Obviously, as shown by this article and silicon lottery, some can, but I have seen no information yet as to what percentage can reach this frequency. Obviously, since the absolute top clockspeed on silicon lottery is 4.25, that range is at the ragged edge for even the best of chips.

4.20GHz-4.30GHz AVX stable is the max for most (without going to risky voltages or liquid cooling). I have seen some results online of 4.35GHz-4.45GHz, but I suspect those aren't the norm.

On the other hand, Guru3D got 4400MHz all core on their 2700X @ 1.4V, for what that's worth:
http://www.guru3d.com/articles_pages/amd_ryzen_7_2700x_review,29.html

 

[coercitiv]

I was talking about all core turbo on the stock 2600x, in reference to scannal's post that the 2600x is faster (I assumed he was talking about stock) than the 8400. If you can show me documentation that the 2600x has a 4.2 ghz all core turbo, I would be most interested to see it.

In games the 2600X behaves like a 4.1Ghz overclocked 2600. In case you're tempted to use the the 8400 perf numbers in this graph, take note they're based on using Z370 & 3200 CL14 RAM.

 

[epsilon84]

In games the 2600X behaves like a 4.1Ghz overclocked 2600. In case you're tempted to use the the 8400 perf numbers in this graph, take note they're based on using Z370 & 3200 CL14 RAM.

And how much do you think an 8400 loses in performance with DDR4-2666? About 5%, HWUB already tested this: https://www.techspot.com/amp/review/1608-core-i5-8400-vs-ryzen-5-1600-best-value/page3.html

So we are pretty much back to performance parity, with the 8400 having the pricing advantage because of DDR4 2666 vs CL14 DDR4 3200, not to mention the 8400 itself is $40 cheaper than a 2600X

 

[f2bnp]

The reason Intel is considered a bit cheaper (if running a 8400/B360) is that you can save a bit of money on DDR4-2666 instead 3200 because the platform is limited to 2666 speeds anyway, so anything faster is a waste. Your example proves exactly that - even with DDR4 3200 it will only run at 2666 speeds on that motherboard.

It's not a huge amount cheaper, I think 2666 is about $20 or $30 cheaper than 3200, and the CPU itself is slightly cheaper too. But the point is rather you can get slightly better gaming performance for slightly *less* money. Which makes Intel the better value for gaming *only* builds. For anything else, like productivity or even a mix of gaming and productivity I would choose the 2600 for the extra threads, even if it costs slightly more.

Like I said, this may be true in some countries, but it certainly isn't true in mine. 2666MHz is at the same price as a kit of 3200MHz 16GB sticks. 2133MHz is maybe 10-15E cheaper overall, amazingly bad value. Hopefully, the prices will go down eventually...

 

[TheELF]

In games the 2600X behaves like a 4.1Ghz overclocked 2600. In case you're tempted to use the the 8400 perf numbers in this graph, take note they're based on using Z370 & 3200 CL14 RAM.

And how much do you think an 8400 loses in performance with DDR4-2666? About 5%, HWUB already tested this: https://www.techspot.com/amp/review/1608-core-i5-8400-vs-ryzen-5-1600-best-value/page3.html

So we are pretty much back to performance parity, with the 8400 having the pricing advantage because of DDR4 2666 vs CL14 DDR4 3200, not to mention the 8400 itself is $40 cheaper than a 2600X

And the i3-8350k is limited to DDR4-2400 at least according to intel...
What is the max OC for the i3?
Sources suggest that 4.8Ghz should be pretty easy.
If gaming is all you are after a intel quad still kills anything else AMD OR Intel in value and is even close enough to the top performance for it to not play any real role anymore,it's ~10% slower in the "kids" resolution of 1080p.
https://www.reddit.com/r/hardware/comments/7ae6f1/silicon_lotterys_i38350k_overclocking_averages/

 

[Ranulf]

Hmm, ok that is not what was said in the video. Either way, low core count for such a CPU intensive game.

Its been that way for almost 8 years now. SC2 benchmarks are not to be taken seriously for anything other than, can it run with more units on screen/in game than before without becoming a total slideshow. Comparing it to any other games is just silly. It has also, like most Blizzard games favored Intel cpus over AMD.

Or, in other words, on the same budget you can either get:
1. 8400 / B360 / DDR4 2666 + GTX 1080
OR
2. 2600+HSF / B350 / DDR4 3200/3466 + GTX 1070

Who on a budget is really going to buy a 1080 or even 1070 these days? Go with Intel for the $80 savings and put it into a good hsf that will be quiet and keep the chip cool prolonging the life of the system. GPUs come and go most of the time, cpus are lasting 4-5 years, even the I-5 level stuff. Though I favor the Ryzen 6/12 for longevity and price/perf.

 

[SPBHM]

price/perf of Intel is clearly ahead for gaming,

you can get the i5 8500 + H310 board for $250, run it with the stock cooler forever, and use the less expensive DDR4 you can find all with default settings and it will perform quite well, ahead of most Ryzen configurations, as good as a stock 7700K basically, that's pretty great...

longevity is certainly a strong argument in favor of Ryzen, and if you are doing video work for example it's the more logical choice, but for a cost effective gaming machine for right now, it's hard to argue against the i5s

 

[AtenRa]

Actualy it is 9% faster. When saying "faster than", one uses the smaller number as the denominator. 145/133 = 1.090, or 9% faster.

Ahh hell my bad, thanks for the correction

 

[AtenRa]

From newegg,

AMD R5 2600X = $210
ASROCK AB350M Pro4 = $60
G.SKILL Ripjaws V Series 16GB (2 x 8GB) DDR4 3600 = $181

Total = $451

Intel Core i5 8400 = $180
Cheapest B360 motherboard MSI B360M PRO-VD = $66
G.SKILL Ripjaws V Series 16GB (2 x 8GB) DDR4 2666 = $167

Total = $413



For $38 more you get,

12x Threads vs 6 Threads,
Within 5% Gaming performance and more than 40-50% on average in multi threaded apps vs Core i5 8400.
Dual M2.0 ports one of which directly from the CPUs 4x PCIe 3.0 lanes vs single M2.0 port from Chipset (B360 through DMI 3.0)
Quad Memory slots vs Dual slots
Upgradable platform up to 2020 (Future 7nm CPUs)
Overclocking enabled.

Even if you are only a gamer and 8400 is currently 5% faster, personally I dont see any reason to chose the Core i5 8400 over the 2600X setup since next Gen games will be more optimized for the RYZEN architecture + better platform features and future upgrade-ability + OC enabled platform.

 

[LTC8K6]

You can probably go with a Z370 and 8600K for ~$100 more than the 8400.

I see Z370 boards for ~$110 and 8600K chips for ~$225.

But we do have this competition again, which is good.

It's kind of neat to be able to search around and pick out comparable systems from both AMD and Intel.

We now need AMD to bring out something that is King of the desktop CPU hill.

 

[epsilon84]

From newegg,

AMD R5 2600X = $210
ASROCK AB350M Pro4 = $60
G.SKILL Ripjaws V Series 16GB (2 x 8GB) DDR4 3600 = $181

Total = $451

Intel Core i5 8400 = $180
Cheapest B360 motherboard MSI B360M PRO-VD = $66
G.SKILL Ripjaws V Series 16GB (2 x 8GB) DDR4 2666 = $167

Total = $413



For $38 more you get,

12x Threads vs 6 Threads,
Within 5% Gaming performance and more than 40-50% on average in multi threaded apps vs Core i5 8400.
Dual M2.0 ports one of which directly from the CPUs 4x PCIe 3.0 lanes vs single M2.0 port from Chipset (B360 through DMI 3.0)
Quad Memory slots vs Dual slots
Upgradable platform up to 2020 (Future 7nm CPUs)
Overclocking enabled.

Even if you are only a gamer and 8400 is currently 5% faster, personally I dont see any reason to chose the Core i5 8400 over the 2600X setup since next Gen games will be more optimized for the RYZEN architecture + better platform features and future upgrade-ability + OC enabled platform.

I share similar thoughts about the benefits of AMD as a platform, by why would you use CL19 DDR4 for AMD when it's been proven that TIGHT timings (CL14) are what gets AMD close to Intel in gaming performance? I'm really confused here. I can assure you with that RAM, AMD will be more like 15% behind in gaming, not 5%. I can understand you're trying to make a point about AMD being better value, but at least choose memory that isn't going to severely cripple Ryzen for the sake of saving $30 compared to a decent set of CL14 B-Die memory.

CL14 3200/3400 >>>>>>>>>>>>> CL 19 3600. Did I use enough arrows to make my point?

Here, maybe this will make more sense than my arrows:

Thats with CL16 3600, btw. Your suggested CL19 3600 is going to run like utter crap on Ryzen, despite the high frequency, because the high latency is what tanks Ryzen gaming performance. I wouldn't be surprised if it performed similarly to the CL16 2933.

 

[IEC]

Don't forget that the speed in MT/s makes a significant difference as fabric/uncore clock is tied to mem clock. In addition, 3600 CL19 typically can be downclocked to 3200 CL16 on Ryzen as the effective CAS latency is essentially the same.

 

[epsilon84]

Don't forget that the speed in MT/s makes a significant difference as fabric/uncore clock is tied to mem clock. In addition, 3600 CL19 typically can be downclocked to 3200 CL16 on Ryzen as the effective CAS latency is essentially the same.

Even so, CL16 3200 is still 'too loose' for Ryzen IMO, look at the figures above, you're losing about 10% in gaming performance compared to Stilt's CL15 timings at 3466, which are generally only possible on B-Die memory.

Not to mention, tweaking memory timings is an extra step and potentially even more daunting than overclocking the CPU itself. I'm a seasoned overclocker and I still have to google memory tweaking articles to see which subtimings offer the most benefit, let alone all the extra stability testing that is involved. Every time a timing is changed, you have to stability test. Its quite a hassle, and not something every gamer will want to go through, even if they are overclockers. Something like Stilts preset would work wonders on Ryzen, and save the hassle of tweaking all the subtimings, but again, only really possibly on quality B-Die memory.

The 8400 really is the 'simple' and 'easy' option to get decent gaming performance, at a good price. It's within 5% of a fully maxed out 2600 platform which costs about $80 more.

Now, honestly, if I was buying for myself, I would get the 2600 and tweak it to my hearts content, I personally find locked CPUs boring as hell, but if I was buying for a family member or friend and just wanted to give them the best bang for buck gaming option without the hassle of having to overclock and tweaking memory timings? I'd just get the 8400/B360/DDR4 2666. Tell them to get a better GPU with the $80 saved, or shout me a dinner as thanks )

 

[coercitiv]

Even so, CL16 3200 is still 'too loose' for Ryzen IMO, look at the figures above, you're losing about 10% in gaming performance compared to Stilt's CL15 timings at 3466, which are generally only possible on B-Die memory.

But AFAIK the Stilt's optimized timings have not been taken into consideration for the purpose of this review and this discussion. If anything, they would bring additional performance uplift over the overclocked results discussed above.

Far Cry 5 results in the OP review:
R5 2400 @ 4.2Ghz with 3400Mhz RAM - 102/118 FPS

Far Cry 5 results in the optimized timings review:
R5 2400 @ 4.2Ghz with 3400Mhz RAM /w Stilt's timings - 108/125 FPS

 

[epsilon84]

But AFAIK the Stilt's optimized timings have not been taken into consideration for the purpose of this review and this discussion. If anything, they would bring additional performance uplift over the overclocked results discussed above.

Far Cry 5 results in the OP review:
R5 2400 @ 4.2Ghz with 3400Mhz RAM - 102/118 FPS

Far Cry 5 results in the optimized timings review:
R5 2400 @ 4.2Ghz with 3400Mhz RAM /w Stilt's timings - 108/125 FPS

https://www.techspot.com/review/1627-core-i5-8400-vs-ryzen-5-2600/

Then for the Ryzen 5 2600 we used the Asus ROG Crosshair VII Hero and we have two configurations: a stock out of the box configuration using the Wraith Stealth box cooler and G.Skill FlareX CL14 memory clocked at 2933 MHz. And a second test setup, running with a 4.2 GHz all-core overclock using aggressively-tuned G.Skill Sniper X DDR4-3400 memory with tightened sub timings. Here the cooler was upgraded to the Corsair H115i Pro.

I can't remember if he specifically mentioned Stilts preset in the actual video, but regardless, those are tuned timings on the video at 3400 speeds, definitely not your run of the mill 'XMP set and forget' settings.

 

[epsilon84]

I'm not the most concise at posting so I apologise if I've gone around in circles a bit in this thread, but the point I'm trying to make is that, for Ryzen at least, everything points to memory latency (and to a lesser extent, frequency) as key to get it competitive with Coffee Lake chips in gaming.

In fact, running suitably tight timings is probably more important than overclocking the CPU itself. Even a stock 2600 will get very close to a 8400 with tuned B Die kits running at 3200 CL14 or 3466 CL15. I would say that the majority of the gains you see in the HWUB video are due to the memory tweaking, not the actual overclock.

 

[Kenmitch]

Viewing the review I guess one could argue for either rig as the best for them depending on the intended use. AMD does have somewhat of an advantage with future processor support in the end.

I'm looking forward to seeing what AMD can bring to the table on the next go around.

 

[coercitiv]

I can't remember if he specifically mentioned Stilts preset in the actual video, but regardless, those are tuned timings on the video at 3400 speeds, definitely not your run of the mill 'XMP set and forget' settings.

Then I must say I'm quite confused by the review results: the original "Stilt timings" review showed considerable gains over 3200 CL14, and the difference does not seem to be increasing when using 2933 CL14 as base in the OP review. Granted there's only one game we can directly compare since Warhammer was tested at different detail settings, and maybe even Farcry data is different.

 

[wahdangun]

You can probably go with a Z370 and 8600K for ~$100 more than the 8400.

I see Z370 boards for ~$110 and 8600K chips for ~$225.

But we do have this competition again, which is good.

It's kind of neat to be able to search around and pick out comparable systems from both AMD and Intel.

We now need AMD to bring out something that is King of the desktop CPU hill.

Yeah, it's refreshing to see this, it's like old days in gpu section but with cpu instead.

King of the Hill for cpu will be served by TR2, can't wait for 16 core 32 thread, 4,4/4,5 turbo, and lot's of PCIe.

 

[wahdangun]

Even so, CL16 3200 is still 'too loose' for Ryzen IMO, look at the figures above, you're losing about 10% in gaming performance compared to Stilt's CL15 timings at 3466, which are generally only possible on B-Die memory.

Not to mention, tweaking memory timings is an extra step and potentially even more daunting than overclocking the CPU itself. I'm a seasoned overclocker and I still have to google memory tweaking articles to see which subtimings offer the most benefit, let alone all the extra stability testing that is involved. Every time a timing is changed, you have to stability test. Its quite a hassle, and not something every gamer will want to go through, even if they are overclockers. Something like Stilts preset would work wonders on Ryzen, and save the hassle of tweaking all the subtimings, but again, only really possibly on quality B-Die memory.

The 8400 really is the 'simple' and 'easy' option to get decent gaming performance, at a good price. It's within 5% of a fully maxed out 2600 platform which costs about $80 more.

Now, honestly, if I was buying for myself, I would get the 2600 and tweak it to my hearts content, I personally find locked CPUs boring as hell, but if I was buying for a family member or friend and just wanted to give them the best bang for buck gaming option without the hassle of having to overclock and tweaking memory timings? I'd just get the 8400/B360/DDR4 2666. Tell them to get a better GPU with the $80 saved, or shout me a dinner as thanks )

But like you said, "simple and easy " is boring, we are enthusiast.

And memory overclocking is easy on Ryzen, even in some low end board like A320, I can oc the ram to 299, it's even have the preset, you just enter bios, select memory speed you desire and save, and what surprised me was asrock A320 have bclk oc in it, so you can increase the boost clock, but never tried it because the stock cooler kinda crap.