• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

MSI GeForce RTX 4070 Gaming X Trio

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
27,053 (3.71/day)
Processor Ryzen 7 5700X
Memory 48 GB
Video Card(s) RTX 4080
Storage 2x HDD RAID 1, 3x M.2 NVMe
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 10 64-bit
The MSI GeForce RTX 4070 Gaming X Trio is the company's factory-overclocked custom design variant. Out of the box it is the fastest RTX 4070 that we've tested, thanks to a large power limit increase paired with the factory OC. Noise levels and temperatures are outstanding, too.

Show full review
 
Joined
Sep 14, 2020
Messages
498 (0.38/day)
Location
Greece
System Name Office / HP Prodesk 490 G3 MT (ex-office)
Processor Intel 13700 (90° limit) / Intel i7-6700
Motherboard Asus TUF Gaming H770 Pro / HP 805F H170
Cooling Noctua NH-U14S / Stock
Memory G. Skill Trident XMP 2x16gb DDR5 6400MHz cl32 / Samsung 2x8gb 2133MHz DDR4
Video Card(s) Asus RTX 3060 Ti Dual OC GDDR6X / Zotac GTX 1650 GDDR6 OC
Storage Samsung 2tb 980 PRO MZ / Samsung SSD 1TB 860 EVO + WD blue HDD 1TB (WD10EZEX)
Display(s) Eizo FlexScan EV2455 - 1920x1200 / Panasonic TX-32LS490E 32'' LED 1920x1080
Case Nanoxia Deep Silence 8 Pro / HP microtower
Audio Device(s) On board
Power Supply Seasonic Prime PX750 / OEM 300W bronze
Mouse MS cheap wired / Logitech cheap wired m90
Keyboard MS cheap wired / HP cheap wired
Software W11 / W7 Pro ->10 Pro
Impressive, but still overpriced. Also the same (unnecessary) length with 4070 ti.
Asus dual looks like the best choice imho.
 
Joined
Feb 20, 2019
Messages
7,310 (3.86/day)
System Name Bragging Rights
Processor Atom Z3735F 1.33GHz
Motherboard It has no markings but it's green
Cooling No, it's a 2.2W processor
Memory 2GB DDR3L-1333
Video Card(s) Gen7 Intel HD (4EU @ 311MHz)
Storage 32GB eMMC and 128GB Sandisk Extreme U3
Display(s) 10" IPS 1280x800 60Hz
Case Veddha T2
Audio Device(s) Apparently, yes
Power Supply Samsung 18W 5V fast-charger
Mouse MX Anywhere 2
Keyboard Logitech MX Keys (not Cherry MX at all)
VR HMD Samsung Oddyssey, not that I'd plug it into this though....
Software W10 21H1, barely
Benchmark Scores I once clocked a Celeron-300A to 564MHz on an Abit BE6 and it scored over 9000.
The fact you can throw 40 more Watts at it (a 20% increase) and only lose 1% efficiency shows you just how artificially hamstrung this card is from Nvidia, just so they don't cannibalise the (very lucrative ) sale of overpriced 4070Ti cards.

We really do need AMD and Intel to hurry up and answer with some competition because these anti-consumer shenanigans are exactly the bad behaviour we've come to expect when Nvidia go uncontested into any market segment.
 
Joined
Apr 13, 2022
Messages
989 (1.32/day)
The fact you can throw 40 more Watts at it (a 20% increase) and only lose 1% efficiency shows you just how artificially hamstrung this card is from Nvidia, just so they don't cannibalise the (very lucrative ) sale of overpriced 4070Ti cards.

We really do need AMD and Intel to hurry up and answer with some competition because these anti-consumer shenanigans are exactly the bad behaviour we've come to expect when Nvidia go uncontested into any market segment.

Won't matter. nvidia won't drop their prices and will still sell the best.
 

tomn2o

New Member
Joined
Apr 13, 2023
Messages
1 (0.00/day)
at reviev there is mistake, it is ventus weight and photo:)
 
Joined
Jul 20, 2020
Messages
829 (0.60/day)
System Name Gamey #1 / #2
Processor Ryzen 7 5800X3D / Core i7-9700F
Motherboard Asrock B450M P4 / Asrock B360M P4
Cooling IDCool SE-226-XT / CM Hyper 212
Memory 32GB 3200 CL16 / 32GB 2666 CL14
Video Card(s) PC 6800 XT / Soyo RTX 2060 Super
Storage 4TB Team MP34 / 512G Tosh RD400+2TB WD3Dblu
Display(s) LG 32GK650F 1440p 144Hz VA
Case Corsair 4000Air / CM N200
Audio Device(s) Dragonfly Black
Power Supply EVGA 650 G3 / Corsair CX550M
Mouse JSCO JNL-101k Noiseless
Keyboard Steelseries Apex 3 TKL
Software Win 10, Throttlestop
The fact you can throw 40 more Watts at it (a 20% increase) and only lose 1% efficiency shows you just how artificially hamstrung this card is from Nvidia, just so they don't cannibalise the (very lucrative ) sale of overpriced 4070Ti cards.

That's not shown anywhere in the review. The 1% lower efficiency is at 9W more power than the Founders 4070 according to this:

 

praescepter

New Member
Joined
Apr 13, 2023
Messages
2 (0.01/day)
The weight shown in the teardown section is not the correct one.
Actually you can see that the pic with the scale is the Ventus 4070.
 
Joined
Feb 20, 2019
Messages
7,310 (3.86/day)
System Name Bragging Rights
Processor Atom Z3735F 1.33GHz
Motherboard It has no markings but it's green
Cooling No, it's a 2.2W processor
Memory 2GB DDR3L-1333
Video Card(s) Gen7 Intel HD (4EU @ 311MHz)
Storage 32GB eMMC and 128GB Sandisk Extreme U3
Display(s) 10" IPS 1280x800 60Hz
Case Veddha T2
Audio Device(s) Apparently, yes
Power Supply Samsung 18W 5V fast-charger
Mouse MX Anywhere 2
Keyboard Logitech MX Keys (not Cherry MX at all)
VR HMD Samsung Oddyssey, not that I'd plug it into this though....
Software W10 21H1, barely
Benchmark Scores I once clocked a Celeron-300A to 564MHz on an Abit BE6 and it scored over 9000.
That's not shown anywhere in the review. The 1% lower efficiency is at 9W more power than the Founders 4070 according to this:
Good point, I was forgetting that power limit != actual TDP
In saying that it's still 3% faster for a 4% power increase. Normally GPUs these days are clocked well beyond the point of diminishing returns.
 
Joined
Mar 9, 2020
Messages
80 (0.05/day)
"Everybody agrees that ray tracing is the future" Didn't they say that about PhysX and Hairworks? Most implementations of Ray Tracing are hardly noticeable whilst actually gaming anyway.
In other words...It's not much more than an expensive gimmick.
 
Joined
Feb 20, 2019
Messages
7,310 (3.86/day)
System Name Bragging Rights
Processor Atom Z3735F 1.33GHz
Motherboard It has no markings but it's green
Cooling No, it's a 2.2W processor
Memory 2GB DDR3L-1333
Video Card(s) Gen7 Intel HD (4EU @ 311MHz)
Storage 32GB eMMC and 128GB Sandisk Extreme U3
Display(s) 10" IPS 1280x800 60Hz
Case Veddha T2
Audio Device(s) Apparently, yes
Power Supply Samsung 18W 5V fast-charger
Mouse MX Anywhere 2
Keyboard Logitech MX Keys (not Cherry MX at all)
VR HMD Samsung Oddyssey, not that I'd plug it into this though....
Software W10 21H1, barely
Benchmark Scores I once clocked a Celeron-300A to 564MHz on an Abit BE6 and it scored over 9000.
"Everybody agrees that ray tracing is the future" Didn't they say that about PhysX and Hairworks? Most implementations of Ray Tracing are hardly noticeable whilst actually gaming anyway.
In other words...It's not much more than an expensive gimmick.
I bought RTX 20-series at launch and I've been continually disappointed by how shit RT is.
Despite that, I agree with @W1zzard; Ray tracing is the future.

I'm thinking 5+ years into the future though; All of the raytracing right now is still an early-adopter's compromise in some way - GPUs simply don't have the performance to raytrace yet.
Look at the 4090 path tracing CP2077. It barely hits 1080p60 before you add DLSS fakery. Meanwhile, half-decade old cards which launched at $300-350 (so 1/5th the cost, 5 years ago) perform the same without raytracing.

Eventually, game devs will focus on RT but for now, 99% of their target audience doesn't have the GPU to handle it. As a result, we currently get RT bolted on as an afterthought. It's half-baked, semi-RT fake-traced mock-up of what a proper path-traced game should be, and even then it halves the FPS for minimal visual gains. This half-assed early RT is just a stepping stone to better things - a necessary step, but not a good one.
 
Last edited:
Joined
Sep 27, 2008
Messages
1,019 (0.18/day)
"Everybody agrees that ray tracing is the future" Didn't they say that about PhysX and Hairworks? Most implementations of Ray Tracing are hardly noticeable whilst actually gaming anyway.
In other words...It's not much more than an expensive gimmick.
The major difference is ray tracing is not proprietary.
 

SFA

New Member
Joined
Apr 16, 2023
Messages
2 (0.01/day)
Great review, as always! But why didn't you include RTX4080 & RTX4090 in the benchmarks? I was very disappointed not to see them. The previous reviews for those are out of date, or rather not directly comparable with the recent ones done for RTX4070 & RTX4070 Ti.

The tests for RTX4070/Ti are saying "TPU Custom Scene", which makes the benchmarks different from what was done on RTX4090. I only noticed the difference after detecting a disturbance in some numbers when I was looking up RTX4090 & 70/Ti results and doing comparisons. I had to rely on a combination of different reviews to achieve a direct comparison at different resolutions, but I ended up pulling my hair out as it was tiring and confusing as heck. I trust TPU and I don't want to look for my needs elsewhere.

Why was I doing that in the first place, you might ask? I was looking for an apples-to-apples comparison because im in a dilemma and I haven't decided on which route to go with yet (either RTX4090+7950X3D@4K, or go wallet-friendly RTX4070/Ti+7800X3D@1440p). I can easily afford the top stuff but im leaning more towards the budget option for obvious reasons, but im also fighting with myself about the importance of going 4K this time (been on 1080p since 2009) because I don't plan on upgrading anytime soon (good monitors are expensive as heck), but then it pains me mentally to pay an insane amount of ~$1866 for RTX4090 which will become obsolete in probably less than a year after the next gen comes out and I see RTX5070 dominating the RTX4090 for less than half the price. There's also the RT stuff in future games which makes the budget option not the best choice here, but then I will be saving close to $1500 in the bank which I could use in 1-2 years to upgrade to the next gen RTX, but then I would have already bought 1440p and there's no way I would also buy a new 4K monitor after just a year or two!

I could literally go for hours with the buts. Going crazy here trying to decide and you're not helping me with your decision to take off RTX4090 from your latest benchmarks :(
 
Last edited:
Joined
Jul 20, 2020
Messages
829 (0.60/day)
System Name Gamey #1 / #2
Processor Ryzen 7 5800X3D / Core i7-9700F
Motherboard Asrock B450M P4 / Asrock B360M P4
Cooling IDCool SE-226-XT / CM Hyper 212
Memory 32GB 3200 CL16 / 32GB 2666 CL14
Video Card(s) PC 6800 XT / Soyo RTX 2060 Super
Storage 4TB Team MP34 / 512G Tosh RD400+2TB WD3Dblu
Display(s) LG 32GK650F 1440p 144Hz VA
Case Corsair 4000Air / CM N200
Audio Device(s) Dragonfly Black
Power Supply EVGA 650 G3 / Corsair CX550M
Mouse JSCO JNL-101k Noiseless
Keyboard Steelseries Apex 3 TKL
Software Win 10, Throttlestop
Great review, as always! But why didn't you include RTX4080 & RTX4090 in the benchmarks? I was very disappointed not to see them. The previous reviews for those are out of date, or rather not directly comparable with the recent ones done for RTX4070 & RTX4070 Ti.

The tests for RTX4070/Ti are saying "TPU Custom Scene", which makes the benchmarks different from what was done on RTX4090. I only noticed the difference after detecting a disturbance in some numbers when I was looking up RTX4090 & 70/Ti results and doing comparisons. I had to rely on a combination of different reviews to achieve a direct comparison at different resolutions, but I ended up pulling my hair out as it was tiring and confusing as heck. I trust TPU and I don't want to look for my needs elsewhere.

It seems that when you need is on this page in the review:


Wizz changed the test system to Windows 11 (with other minor changes) early this year so those late 2022 tests in the original reviews are accurate for that Win10 system but can't be directly compared here.

But no matter, just have a look at the link above for current coverage from the 4090 all the way down to the 6500XT.
 

SFA

New Member
Joined
Apr 16, 2023
Messages
2 (0.01/day)
It seems that when you need is on this page in the review:


Wizz changed the test system to Windows 11 (with other minor changes) early this year so those late 2022 tests in the original reviews are accurate for that Win10 system but can't be directly compared here.

But no matter, just have a look at the link above for current coverage from the 4090 all the way down to the 6500XT.
Oh, I didn't even notice the change from Win 10 to 11! All I noticed was that "TPU Custom Scene" on some of the reviews, and it made me think he started testing using his own custom scenario (which wasn't applied on the previous reviews, so that makes them sort of obsolete).

This means I have to look for the most recent RTX4070 review to be able to compare it to RTX4070 Ti & RTX4080 & RTX4090, but sadly 80/90 were not included in the recent review.

Any advice? I don't really care for that relative performance stuff, I need to look at actual up-to-date numbers.

Also, here's something that looks interesting:


1681649466551.png
1681649472782.png

1681649484888.png
1681649491737.png


What's going on here? Look at the RTX4090 results (185.4 vs 211.3 FPS).

The difference of (CPU power limit, MB, thermal paste, PSU, case, OS, drivers) should NOT show this huge +/-12% difference. This is like going from RTX4090 to RTX4090 Ti (which doesn't exist), or vice-versa.

Also, shouldn't the new updated test setup (on the left) with Win 11 + 320W CPU limit + newer drivers = give out a higher performance? Why on earth would the newer and better be 12% slower? This is depressing because here I am thinking I'll go all out with my new build and finally move to Win 11.
 
Last edited:
Top