3800x vs 9900k

[Robert896r1]

for a 5GHZ AVX clock I needed one of these, the AIO had to go in the bin :

Spoiler

I am very happy with it as well.

Yeah that's a beast. I'm on a lowly Alphacool 360LT which can handle about 150a/230w in normal conditions and goes upto 165a/255w on a very cold night like last night with the window open. Basically cheating ambient.

My 9900k scales well. No avx offset on below:
50/47x = 1.30v
51/47x = 1.35v
52/47x = 1.40v

All those tested with OCCT Large + AVX2 2 hours and 10-30 loops of x264 stress test. For the 52/47x I had to open the windows to keep from temp throttling. I'm daily at the 5.1 settings.

 

[Robert896r1]

3dmark 5.1GHz 11573 (3000 RAM) 64-bit Windows 10 (10.0.17763)

https://www.3dmark.com/spy/8140427

That's a low Timespy score for a 9900k. I wonder if he was throttling or the poor ram is holding him back.

very high 12's or low 13's is normal for 9900k with decent ram. https://www.3dmark.com/3dm/33358882 < mine before ram tuning I think.

 

[zx128k]

Yeah that's a beast. I'm on a lowly Alphacool 360LT which can handle about 150a/230w in normal conditions and goes upto 165a/255w on a very cold night like last night with the window open. Basically cheating ambient.

My 9900k scales well. No avx offset on below:
50/47x = 1.30v
51/47x = 1.35v
52/47x = 1.40v

All those tested with OCCT Large + AVX2 2 hours and 10-30 loops of x264 stress test. For the 52/47x I had to open the windows to keep from temp throttling. I'm daily at the 5.1 settings.

Prime 95 fft 8k avx and watch what happens.

 

[zx128k]

That's a low Timespy score for a 9900k. I wonder if he was throttling or the poor ram is holding him back.

very high 12's or low 13's is normal for 9900k with decent ram. https://www.3dmark.com/3dm/33358882 < mine before ram tuning I think.

Stock RAM for the 9900k is 2666, not the 3600 ram you have. The ram speed increase is bigger than the clock speed. https://www.intel.co.uk/content/www/uk/en/products/processors/core/i9-processors/i9-9900k.html You mean an overclocked 9900k, with a RAM overclock and not a stock 9900k which is not that fast compared to a 3800x. Both being near to equal.

Some 9900k, are able to overclock far higher than the 3800x in performance, that's the difference. A 5.1GHz overclock is top 5% of chips. 95% of 9900k's can't reach your speeds and your build is hugely more expensive than a 3800x build. So saying 9900k faster than the 3800x because your 9900k is a outlier, is a flawed argument.

https://ibb.co/0n0954T

 

[Robert896r1]

Stock RAM for the 9900k is 2666. https://www.intel.co.uk/content/www/uk/en/products/processors/core/i9-processors/i9-9900k.html You mean an overclocked 9900k, with a RAM overclock and not a stock 9900k which is not that fast compared to a 3800x. Both being near to equal.

I'm confused. You linked an overclocked Timespy score?

People who buy a 9900k are not buying cheapest ram possible. They are also likely to be enthusiast class consumers inclined to all core OC out of their chips. One platform having OC headroom over another isn't the fault of that platform.

The benefits of a 3800x over a 9900k are the cost savings (lower MSRP + free cooler) and overall lower cooling requirements. The 3800x is a great chip and awesome value but it doesn't have the pure performance of a 9900k in part because the 9900k does have headroom built in.

 

[sideways14a]

The 3800x though does have performance that the 9900k struggles with sometimes though.
For instance its widely known AMDs SMT implementation is better, again helping when the CPU is loaded.

The 9900k is only fast in certain circumstances, so is the 3800x.

 

[zx128k]

I'm confused. You linked an overclocked Timespy score?

People who buy a 9900k are not buying cheapest ram possible. They are also likely to be enthusiast class consumers inclined to all core OC out of their chips. One platform having OC headroom over another isn't the fault of that platform.

The benefits of a 3800x over a 9900k are the cost savings (lower MSRP + free cooler) and overall lower cooling requirements. The 3800x is a great chip and awesome value but it doesn't have the pure performance of a 9900k precisely because the 9900k does have headroom built in.

What I am showing is the 3800x is able to match the 9900k for performance within reason. Apart from the top 30% of 9900k the 3800x is not slower.

This is what an average 9900k system looks like, https://www.3dmark.com/spy/8230671 which is 4.9GHz 3400 RAM (overclocked from 3200) and CPU points 11440. Prue clock speed does not put the 9900k ahead, a RAM overclock is needed. With the 3800x you hit the hard wall at 3800MHz IF 1900. The 9900k you could keep going if the chip lets you.

This is what happens if you take the RAM overclock away.

https://www.3dmark.com/spy/8034395 which is 5.0GHz 3,008 RAM and CPU points 10913.

https://www.3dmark.com/spy/8696801 which is 4.9GHz 2,666 MHz RAM and CPU points 10041. This is a 9900k at all cores 4.9GHz with stock RAM speed. Remember that £100-150 AIO you need.

https://www.3dmark.com/spy/8633001 which is 4.8GHz 2,138 MHz RAM and CPU points 10041.

Stock 3800x is faster than the 9900k clocked at 4.9GHz all cores and below with stock RAM.

Stock 9900k https://www.3dmark.com/spy/8317919 with 3600 RAM and CPU points 9782.

If you spend a fortune on your 9900k build you can become faster than the 3800x. Most builds won't be faster, that's the point. This is why Intel needs 5GHz all core fast. To be ahead they need 5GHz all cores and fast RAM.

This is likely to be a decent comparison to the 3800x I have. Remember the price difference.

https://www.3dmark.com/spy/8175621 which is 9900k 5GHz 3200 RAM (£110) and CPU point 11140. 3800x I have is 11300 approx.

It's only the top 9900k systems that are faster than the 3800x in 3d mark, all the rest are on power with the 3800x overclocked as AMD advised.

 

[Robert896r1]

What I am showing is the 3800x is able to match the 9900k for performance within reason. Apart from the top 30% of 9900k the 3800x is not slower.

This is what an average 9900k system looks like, https://www.3dmark.com/spy/8230671 which is 4.9GHz 3400 RAM and CPU points 11440. Prue clock speed does not put the 9900k ahead, a RAM overclock is needed.

This is what happens if you take the RAM overclock away.

https://www.3dmark.com/spy/8230671 which is 5.2GHz 2400 RAM and CPU points 10950.

https://www.3dmark.com/spy/8633001 which is 4.8GHz 2,138 MHz RAM and CPU points 10041.

If you spend a fortune on your 9900k build you can become faster than the 3800x. Most builds won't be faster, that's the point. This is why Intel needs 5GHz all core fast.

Your two links at the bottom are 4.8 vs 4.9 btw, not 5.2.

It's not reasonable to assume that anyone buying a 9900k is buying bottom of the barrel ram. It's a worth "what if" scenario but not a realistic one. I'd say the same for a 3800x owner who would suffer greatly from running ram that low also. It's not limited to intel so I'm not sure how this is a selling point.

I don't believe my Z390 Pro board is any more expensive than a X570 pro. RAM is interchangeable. You do need a good cooler for the 9900k which is a notable cost you can avoid on the 3800x. Again this "fortune" is limited to the MSRP delta and the cost of a cooler.

 

[zx128k]

Your two links at the bottom are 4.8 vs 4.9 btw, not 5.2.

It's not reasonable to assume that anyone buying a 9900k is buying bottom of the barrel ram. It's a worth "what if" scenario but not a realistic one. I'd say the same for a 3800x owner who would suffer greatly from running ram that low also. It's not limited to intel so I'm not sure how this is a selling point.

I don't believe my Z390 Pro board is any more expensive than a X570 pro. RAM is interchangeable. You do need a good cooler for the 9900k which is a notable cost you can avoid on the 3800x. Again this "fortune" is limited to the MSRP delta and the cost of a cooler.

You have to remember that the 9900k build is more expensive and this is without the faster RAM. The 3800x build can be cheap by £100's and still have the best RAM. AMD's market share is going to grow massively. Notice how the reviewers are playing down the true worth of the 3800x. Intel needs a 9900k at 5GHz all cores for the same price as the 9900k for the price to hold. Even then without faster RAM the 3800x will be faster with an IF 1800 and 3600 ram.

 

[zx128k]

I'm a little confused by the comments about pricing.
In the OP, the 3800 system appears to only have 8GB of ram, so there's no wonder that it's cheaper.
Give them the same level of ram, and the CPU cost saving is more than made up for by the additional cost for the motherboard.

Dual kit so two 8GB ram sticks in the kit for 16GB.

 

[zx128k]

I'm confused. You linked an overclocked Timespy score?

People who buy a 9900k are not buying cheapest ram possible. They are also likely to be enthusiast class consumers inclined to all core OC out of their chips. One platform having OC headroom over another isn't the fault of that platform.

The benefits of a 3800x over a 9900k are the cost savings (lower MSRP + free cooler) and overall lower cooling requirements. The 3800x is a great chip and awesome value but it doesn't have the pure performance of a 9900k in part because the 9900k does have headroom built in.

It's the faster RAM, the 9900k is just the same as the 3800x with ram overclocking. With the stock RAM speed an overclocked 5GHz 9900k is nothing to write home about. Remember how few 9900k's get to 5GHz. The 3800x will reach an IF of 1800 and a RAM speed of 3600. With RAM tuning that will hit performance above a stock RAM 5GHz all core 9900k. 9900k approx. 10913 vs. 3800x stock core IF 1800 RAM 3600 approx. >11k.

It's just the 9900k can keep overclocking the RAM to well above 4000MHz fast timings. At equal cost a overclocked 3800x will win. The more you spend on ram and cooling the better the 9900k will be and the more it will pull ahead of the 3800x. There is a point in the spectrum were they are both equal.

 

[Grim5]

You have to remember that the 9900k build is more expensive and this is without the faster RAM. The 3800x build can be cheap by £100's and still have the best RAM. AMD's market share is going to grow massively. Notice how the reviewers are playing down the true worth of the 3800x. Intel needs a 9900k at 5GHz all cores for the same price as the 9900k for the price to hold. Even then without faster RAM the 3800x will be faster with an IF 1800 and 3600 ram.

For the longest time I was using cheap low clocked ddr4 ram on my 8700k. Then I found out how good fast RAM is, I went from 2400mhz CL16 to 3200mhz CL14 and my CPU scores improved by 12%

 

[zx128k]

For the longest time I was using cheap low clocked ddr4 ram on my 8700k. Then I found out how good fast RAM is, I went from 2400mhz CL16 to 3200mhz CL14 and my CPU scores improved by 12%

Really is a big part of system performance. Remember this https://www.youtube.com/watch?v=D_Yt4vSZKVk

 

[Besty]

Some 9900k, are able to overclock far higher than the 3800x in performance, that's the difference. A 5.1GHz overclock is top 5% of chips. 95% of 9900k's can't reach your speeds and your build is hugely more expensive than a 3800x build. So saying 9900k faster than the 3800x because your 9900k is a outlier, is a flawed argument.

hah well, many of us on here are old gits mate, we came here to overclockers in 1999 to buy binned AMD k7s with delta fans which were so loud your ears bled.

20 years on, on this forum at least, we can’t use the outlier argument on this very basis as the people on here are outliers by default, by virtue of being here.

Both AMD and Intel have wild variations in silicon quality, as core counts go up,
binning will be more important.

Should Intel have binned the 9900k better? Yes.

Does it make much of a difference to those here, in the know? not really.

This battle won’t last long anyway as we have yet another round of chips
and motherboards heading our way, get your wallets ready

 

[zx128k]

hah well, many of us on here are old gits mate, we came here to overclockers in 1999 to buy binned AMD k7s with delta fans which were so loud your ears bled.

20 years on, on this forum at least, we can’t use the outlier argument on this very basis as the people on here are outliers by default, by virtue of being here.

Both AMD and Intel have wild variations in silicon quality, as core counts go up,
binning will be more important.

Should Intel have binned the 9900k better? Yes.

Does it make much of a difference to those here, in the know? not really.

This battle won’t last long anyway as we have yet another round of chips
and motherboards heading our way, get your wallets ready

Thank to AMD Intel cut there prices in half. Anyway what I was saying was an item outside of +2σ of the standard deviation of a normal distribution cannot be an example of the mean +/- 2σ normal distribution https://en.wikipedia.org/wiki/Standard_deviation#/media/File:Standard_deviation_diagram.svg

 

[zx128k]

https://www.overclock3d.net/reviews..._7_3700x_ryzen_9_3900x_x470_vs_x570_review/28








Metro exodus benchmark

9900k - 101FPS

3800x OC
Low (CPU test I would pick low) - 166FPS
Normal - 129FPS
High - 101FPS
Ultra - 84

Going higher its more about 2080 vs 2080 ti. Really at 2080 ti should beat me at every setting.

I used low on the benchmark.

Test system from the website.
AMD Ryzen 9 3900X
AMD Ryzen 7 3700X
ASUS ROG X570 Crosshair VIII Hero
ASUS ROG X470 Crosshair VII Hero
AMD Ryzen 9 3900X
G.Skill Trident Z 3600 MHz
Corsair MP500 M.2 - OS Drive
Aorus PCIE4 NVME - Storage Speed Tests
Nvidia RTX 2080Ti
Corsair HX1000i
Corsair H150i
Corsair ML Fans

My system
AMD Ryzen 7 3800x PBO auto OC +200MHz (PBO limits auto) Scalar x1
ASRock X570 Taichi
NVIDIA GeForce RTX 2080
Team Group Inc. 8192 MB (DDR4-2400) - XMP 2.0 - P/N: TEAMGROUP-UD4-3600
(16-15-15-31-46-1 (tCAS-tRC-tRP-tRAS-tCS-tCR) 1899.6 MHz (DDR4-3800) Uncore: 1899.6 MHz)
Full custom water loop 360 rad and full copper block. push/pull fans.
Corsair AX1000

Looks like a stock 9900k does not do well in this game benchmark even with 4400MHz RAM.

 

[zx128k]

Mainboard
ASRock Z390 Taichi Ultimate - Review

Processor
Core i9 9900K (8c/16t) @ defaults - Review

Graphics Cards
GeForce RTX 2080 Super 8GB GDDR6 (NVIDIA Founder/reference edition)

Memory
32 GB (4x 8MB) 3200 MHz DDR4 (Cofee lake platform)
Power Supply Unit
1,200 Watts Platinum Certified Corsair AX1200i - Review

9900k stock with 3200MHz RAM
Ultra 1080p
Metro Exodus
86 FPS
https://www.guru3d.com/articles_pages/geforce_rtx_2080_super_review,12.html


System Spec

• Core i9 9900K
• Z390 (ASRock Tachi Ultimate)
• 16 GB DDR4 3200 MHz CL16
• NVMe M.2. SSD WD Black

9900k stock with 3200MHz RAM
Ultra 1080p
Metro Exodus
79 FPS
https://www.guru3d.com/articles_pages/metro_exodus_pc_graphics_performance_benchmarks,6.html

 

[zx128k]

https://nl.hardware.info/artikel/90...kaart-overleeft-de-post-apocalyptische-wereld

• Intel Core i9 9900K @ 5 GHz (8 cores & 8 threads)
• Gigabyte Z390 Aorus Master moederbord
• G.Skill Trident Z 2x8 GB DDR4-3200 CL14 geheugen
• Seasonic Prime Titanium 1200W voeding
• 2x Samsung 970 Evo 1TB ssd
• NZXT Kraken X62 processorkoeler
• Windows 10 Pro x64 (1809)

Ultra 1080p
Metro Exodus
81 FPS

HT is not a big deal, same website.
https://translate.google.com/transl...9-9900k-in-games-met-en-zonder-ht&prev=search

 

[zx128k]

If someone can do some 9900K benchmarks with a 2080. Low settings would be nice.

 

[Rroff]

HT is not a big deal, same website.
https://translate.google.com/transl...9-9900k-in-games-met-en-zonder-ht&prev=search

Very few games use more than 8 threads currently - having HT off with at least 8 physical cores will probably net you around +5% in games overall (though that article puts it a little lower than that) due to the slight overhead of having SMT on.

Where things can get a bit different is on 4 or 6 core CPUs where turning off HT can result in a reduction in smoothness, even at the same framerate, if you are gaming above 60FPS/Hz with V-Sync disabled.