Exclusive | Nvidia gave laptop makers a useless overblown GeForce MX450 3DMark score that has proven impossible to reach
(June 23, 2021 update: Nvidia has responded with the following statement about the 3DMark score:
"The Firestrike score claimed in this article did not originate from NVIDIA, and does not reflect the performance of the MX450"
Our original article remains below.)
Whenever a new processor launches, chipmakers like to show off their own in-house benchmark numbers and comparisons. These numbers are usually a bit higher than retail results since many are performed under optimal conditions with developer tools and without any overhead from laptop manufacturers. Sometimes, however, these chipmaker numbers can be absurdly high to the point where they can be misleading.
When Nvidia launched its GeForce MX450 series in early 2020, the chipmaker claimed a 3DMark Fire Strike Graphics and Time Spy Graphics score of 7498 and 1677 points, respectively. The Fire Strike score in particular caught our attention as this would rank the GeForce MX450 to be 25 percent higher than even a GeForce GTX 1050 laptop. These specific scores from Nvidia were communicated to laptop manufacturers only according to our laptop manufacturer source.
In reality, however, the Fire Strike score would turn out to be nothing but false in real-world testing conditions. After personally testing 7 different retail laptops each with GeForce MX450 graphics, all of them would score significantly lower in the same benchmark run by Nvidia. The MX450 GPUs in our tested laptops would be about 40 percent slower than Nvidia's in-house developer unit as shown by the comparison charts below. If end-users can't even get close to the performance numbers Nvidia are claiming, then those higher numbers from the chipmaker are arguably worthless.
The strangest part is that the DX12-based Time Spy score provided by Nvidia is actually attainable with the retail models we tested. Both our Lenovo ThinkPad T14 and Asus Zenbook 14 UX435EG retail units, for example, are within just 1 percent of Nvidia's Time Spy numbers. It appears that only the Fire Strike score from Nvidia has been significantly exaggerated.
We had reached out to Nvidia before publishing on the discrepancy, but they have yet to respond to our queries. The takeaway message here is that users should not expect the GeForce MX450 to outperform or even perform similarly to a GeForce GTX 1050 laptop in most cases despite Nvidia's initial claims.
3DMark | |
1920x1080 Fire Strike Graphics | |
Average NVIDIA GeForce GTX 1050 Mobile (4935 - 6651, n=43) | |
HP Pavilion 15-eg | |
Lenovo ThinkPad E15 Gen2-20TD002MGE | |
HP Envy 13-ba1475ng | |
Lenovo ThinkPad T15 Gen2-20W5S00100 | |
Xiaomi RedmiBook Pro 15 | |
Asus Zenbook 14 UX435EG | |
Lenovo ThinkPad T14 Gen2-20W1S00000 | |
2560x1440 Time Spy Graphics | |
Lenovo ThinkPad E15 Gen2-20TD002MGE | |
HP Pavilion 15-eg | |
Asus Zenbook 14 UX435EG | |
MX450 Pre-Sample | |
Average NVIDIA GeForce GTX 1050 Mobile (1416 - 2130, n=11) | |
Lenovo ThinkPad T14 Gen2-20W1S00000 | |
Lenovo ThinkPad T15 Gen2-20W5S00100 |
3DMark 11 - 1280x720 Performance GPU | |
MX450 Pre-Sample | |
Lenovo ThinkPad E15 Gen2-20TD002MGE | |
HP Pavilion 15-eg | |
Xiaomi RedmiBook Pro 15 | |
Average NVIDIA GeForce GTX 1050 Mobile (6744 - 8903, n=42) | |
Lenovo ThinkPad T15 Gen2-20W5S00100 | |
Asus Zenbook 14 UX435EG | |
Lenovo ThinkPad T14 Gen2-20W1S00000 |