The RTX 4060 & RX 7600 Might be the Worst Performing Budget GPUs in Years
NVIDIA announced its RTX 4060 family of Ada Lovelace GPUs just a few days ago. While team green has already made its mark, we are yet to see AMD’s response. With just a few days left before the big day, AMD must be revising the pricing structure for its budget Navi33 offering, but it’s not as if that will matter. Moreover, the RTX 4060’s performance is bound to disappoint you. Here’s why:
RTX 4060, Slower Than the RTX 3060 Ti
The RTX 3060, for the unaware, was able to match an RTX 2070 from Turing. The precedent set by NVIDIA was as follows, the 60-class next-gen GPU offers performance similar to last-gen’s 70-class GPU. Meanwhile, if you fancy yourself a 60-Ti GPU (3060 Ti), you’ll get performance levels equivalent to an 80-class GPU (2080).
To be quite frank, the same was expected of Lovelace, until a few last-minute changes of specifications. The RTX 4070, once had similar specs to the RTX 4070 Ti. Similarly, the RTX 4060 Ti was downgraded from 36 SMs to 34 SMs. To make matters worse, here’s a list of the actual GPU die used by various 60-class GPUs across the years:
- GTX 660: GK106
- GTX 760: GK104
- GTX 960: GM206
- GTX 1060: GP106
- RTX 2060: TU106
- RTX 3060: GA106
- RTX 4060: AD107
Pick the odd one out. It’s as clear as daylight that NVIDIA has downgraded the RTX 4060 from AD106 to AD107. That makes a lot of difference. As an example, the RTX 3060 is based on GA106, and the RTX 3050 is based on GA107. This small downgrade makes the RTX 4060 lose a lot of performance, speaking of which;
……What about the performance?
RTX 4060 Ti’s Performance
The charts shared by NVIDIA do reflect comprehensible and user-friendly numbers, which was absolutely not the case when the RTX 4090 was marketed as 4x faster than the RTX 3090. So, how do we know the ‘Real FPS’ this time around? Thankfully, YouTuber I’m a Mac has done a comprehensive analysis of both the RTX 4060 family and the RX 7600. You can check out his full video here.
Let’s start off with the RTX 4060 Ti, whose 8GB variant has been used for these tests. On average, the RTX 4060 Ti is 13% faster than the RTX 3060 Ti. That is totally overshadowed by the RTX 3060 Ti’s on-average 24% higher performance than the RTX 2060 Super.
We can conclude that the RTX 4060 Ti will roughly perform similarly to the RTX 3070, being 13-15% faster than the 3060 Ti. Do note that ‘For a 1440p gaming experience‘ you ought to choose the 16GB variant which could be faster than the 3070 in certain scenarios.
However, if you are fine with 1080p gaming ‘for the low, low price of $400‘, the RTX 4060 Ti will be the perfect match.
RTX 4060’s Performance
The RTX 4060 does somewhat better than the 4060 Ti, but the overall result will be much more disappointing, as you’ll see in a bit. On average, we see that the RTX 4060 is 18% faster than the RTX 3060, putting it roughly on par with the vanilla RTX 2080. Impressive?…….Not really. The RTX 3060 Ti would still be 7-8% faster than the RTX 4060, putting Ada to shame.
I’m a Mac then combined all these results into one chart to show us a clearer picture. The RTX 4060 fails to match the RTX 3060 Ti, in every game. Do note that FG or Frame Generation is not taken into account here. Frame Generation, as per I’m a Mac is merely a frame smoothing technology. It does not improve latency, which is the sole purpose of a higher framerate.
RDNA3 to The Rescue?
Team Red has not stepped into the battlefield yet, at least formally so we’ll have to judge the performance metrics by AMD’s planned specs. The RX 7600 features the full-fat Navi33 XL GPU packed with 32 CUs, similar to the RX 6600 XT.
|SKU||Chip||CU||Max Clock||Cache||Memory Bus||VRAM||Memory Spec||Speed (Gbps)||TDP|
RDNA3 was tipped to compete fiercely against Ada, though that speculation still remains mere speculation. The efficiency problem with Navi31 and Navi32 was clear. Use a chiplet design, and face a power penalty. This penalty was enough to completely destroy AMD’s efficiency lead against NVIDIA.
Put it simply, transferring data across multiple chiplets would always consume more power than a monolithic design. The main advantage was a cheaper price for the consumer, which is still somewhat debatable.
Navi33, however, is not prone to any of these issues and could show the full power and efficiency of RDNA3. A direct comparison against the RX 6650 XT will be an almost apples-to-apples comparison to show the generation uplift in performance and efficiency.
Spoiler Alert, It Gets Bad Real Quick
Back to the topic, the use of N6 (6nm) has sparked some controversy. In any case, the GPU should be cheaper to make right? Well, AMD is allegedly planning an MSRP Of $299, same as the RTX 4060.
The 7600 is expected to offer performance similar to the RX 6700 (XT/Non-XT) for 300 bucks. The real bummer is the inclusion of just 8GB of VRAM which makes the 7600 a poor choice when compared against the 6700 XT.
The limited frame buffer is a real kick in the teeth for the budget gamer, given how recent AAA titles are keen on devouring your GPU’s memory. A modest price of $199 or even $249 might give some breathing room to the average PC gamer, looking for an upgrade. But that remains to be seen.
The Real Winner
It is obvious, isn’t it? The real winners of this debacle are AMD and NVIDIA. I’m a Mac states that NVIDIA announced the RTX 4060 two months earlier for a specific reason. NVIDIA would price the RTX 4060 at $299, giving AMD adequate space for a slot-in strategy.
Its simple, really, NVIDIA will price their GPUs at a specific MSRP, setting a pricing model for the entire GPU market. AMD just has to ‘slot in’ or price their GPUs much like NVIDIA or a bit lower. Then again, AMD has responded with RDNA2, but that doesn’t seem to be the case with this generation.