3DMark Speed Way Battle: ASUS P9X79 Pro (2011) vs ASUS ROG STRIX X870E-E GAMING WIFI (2024)

[FirstEver.eu]

Recently, I asked a guy with an identical ASUS TUF GeForce RTX 4070 Ti card to check what result he would get in 3DMark Speed Way.
The difference was that my PC is a 13-year-old platform (ASUS P9X79 Pro) with an 11-year-old Xeon E5-1680 v2, and he had a new Ryzen 7 9800X3D and ASUS ROG Strix X870E-E Gaming WiFi board.
Interestingly, his result was far behind (5848 points), until I had to optimize his system myself, then his PC lost by only 0.5 FPS (6055 points).


3DMark Speed Way results:

• Old Xeon E5-1680 v2 OC- 6101: https://www.3dmark.com/3dm/124744332
• New Ryzen 7 9800X3D OC - 6055: https://www.3dmark.com/3dm/124394217

My full specs (I know… it is overkill but I like to upgrade this platform)

MB: ASUS P9X79 Pro (NVMe & ReBAR Mod)
CPU: Intel Xeon E5-1680 V2 @ 4.375GHz (1.36V) 35x125MHz
RAM: 32GB (4x8GB) Corsair Vengeance Quad-DDR3 2400@2333MHz CL10
Cooler: Liquid Freezer II 280
GFX: ASUS GeForce RTX 4070 Ti OC @ 3000/12250MHz
PSU: Cooler Master Silent Pro M2 1500W
SSD: Samsung SSD 990 Pro 4TB / Samsung 870 QVO 8TB
HDD: Seagate ST18000NM000J-2TS 18TB
WiFI: Intel BE200 Wi-Fi 7 / Bluetooth 5.4 PCIe Card
Other: USB 3.2 Gen2 10Gbps card
Monitor: ASUS ROG Swift OLED PG48UQ
Logitech: G900 , G915, G27, Z906, G51, Brio, other...

More about my P9X79 Pro ReBAR mod here.

Today I will upgrade my RAM to 4x8GB G.Skill DDR3 PC3-21300 2666MHz TridentX CL12 (12-13-13-35) (2x F3-2666C12D-16GTXD Kits) and I wonder if I will be able to run them in quad mode like the current Corsair Vengeance 2400@2333MHz CL10. I am currently getting over 70.6GB/s and if I could boost it to 2666MHz I should be getting around 80GB/s, and this will be impressive for DDR3.

​

Cyberpunk 2077: Phantom Liberty (4K Ray-Tracing: Ultra)​

Average FPS 98.62 / Min FPS 89.62 / Max FPS 106.54

If you have any suggestions for optimizing DDR3 memory or BIOS settings, please let us know.

 

[EarthDog]

UL's Speedway is not CPU dependent at all. Makes sense the scores are similar. It's also a synthetic benchmark and you'll see larger differences in actual games... especially at 1080/1440p where the CPU matters more (than at 4K). That video you threw in at the end describes as much I'd imagine (did not watch) as there's no way a 10+ year old CPU hangs with a current-gen model at lower resolutions. 8c processor that runs 3-3.9 GHz just can't hang with 5-5.7.

Today I will upgrade my RAM to 4x8GB G.Skill DDR3 PC3-21300 2666MHz TridentX CL12 (12-13-13-35) (2x F3-2666C12D-16GTXD Kits)

That....... seems like a waste. If I understand right, you're literally not gaining anything. What you gain in the move from 2400->2666, you're losing with the higher CL/timings. Save your money.

and I wonder if I will be able to run them in quad mode like the current Corsair Vengeance 2400@2333MHz CL10.

Why wouldn't you? Four sticks = quad channel on that platform. So long as they are in the right slots, they'll run quad.

 

[FirstEver.eu]

UL's Speedway is not CPU dependent at all. Makes sense the scores are similar. It's also a synthetic benchmark and you'll see larger differences in actual games... especially at 1080/1440p where the CPU matters more (than at 4K). That video you threw in at the end describes as much I'd imagine (did not watch) as there's no way a 10+ year old CPU hangs with a current-gen model at lower resolutions. 8c processor that runs 3-3.9 GHz just can't hang with 5-5.7.

My Xeon is overclocked to 4.375GHz (35x125MHz) and I know that in 1080p and 1440p or lower details the difference can be up to 30-40%, but in the highest details and 4K the games still run decently - for example 100 FPS in Cyberpunk 2077 (RT Ultra).

That....... seems like a waste. If I understand right, you're literally not gaining anything. What you gain in the move from 2400->2666, you're losing with the higher CL/timings. Save your money.

I noticed a big performance increase after changing the FSB from 100MHz -> 125MHz, therefore my DDR3-2400 only works as DDR3-2333.
With my current setup I was in the top 100 with RTX 4070 Ti, and no one else had such an old setup as me :-D
I thought that by upgrading Quad DDR3-2333 -> DDR3-2666 I could get 0.5 additional FPS.

However, something strange happened... performance on DDR3-2666 dropped drastically by 42-48%!

Corsair Vengeance DDR3-2400@2333 CL10-12-12-31



G.SKILL TridentX DDR3-2666@2666 CL12-13-13-35



After activating the XMP profile on G.SKILL, the processor bus is set to 125MHz and the memory to 2666MHz (2x1333MHz)
I couldn't get them to work on the 100MHz bus (even with a lower clock speed, e.g. 2400MHz).
I thought it was a bug in AIDA64, so I decided to add tests in SiSoft Sandra and Hyper PI.



It seems that Quad-DDR3 does nothing for me in games, because the results in Cyberpunk 2077 and 3DMark Speed Way were almost identical.





Does anyone know why memory performance dropped so much?

 

[EarthDog]

but in the highest details and 4K the games still run decently - for example 100 FPS in Cyberpunk 2077 (RT Ultra).

Right. I get that, but there's still a glass ceiling on a decade+ processor. I"m not saying to upgrade or anything, just that it exists. So, it's awesome you're getting 100 FPS... but maybe 110-120 is in the cards with a modern system.

I could get 0.5 additional FPS.

lol, maybe. But not something I'd spend a penny on.

It seems that Quad-DDR3 does nothing for me in games, because the results in Cyberpunk 2077 and 3DMark Speed Way were almost identical.

Normal.

 

[FirstEver.eu]

Ok, not 0.5FPS but I need 1.9FPS to get back to Top 100 in 3DMark Speed Way... there are really fast PCs on the list


Update...
I found a review on TechPowerUp and after a few minutes I managed to reach 78GB/s