RTX 4000 Specs & Price Revealed And It's Not Pretty
Introduction
Nvidia recently held their yearly GTC Conference on September 20, 2022 and shocked the world in more ways than one. They started off with their Nvidia Racer RTX video which looks very impressive. I loved the realistic lighting and post processing. Moving on from 3000 series “Ampere”, Nvidia’s 4000 series is named “Ada Lovelace” who is credited as being the first female programmer. The RTX 4090 and 4080 will definitely bring some serious performance, but there are a few issues that I must discuss. I have been very busy lately and I finally got a chance to sit down and view the GTC Conference several days after the conference. I received a message from a friend and he spoke about an RTX 4080 “Lite”. Initially I thought he was talking about the RTX 4080 initial release that would be followed up by the “Ti” version. After checking out the GTC Conference and the RTX 4000 series price and specifications I knew exactly what he meant. I will save the in-depth micro-architecture features for a RTX 4000 series review article. This article will focus on my issues that I am having with the 4000 series compared to the 3000 series and possibly Nvidia’s reasoning behind their decisions.
Pricing
It should come to no surprise that the RTX 4000 series would be more expensive. The economy appears to be in a recession and stock markets are crashing all over the place. The market was settled on the RTX 3080 selling between $1,200 - $1,800 easily between Q4 2020 to Q2 2022. The RTX 3080 “Ti” release (Q2 2021) didn’t do much to settle prices either, but luckily now that RTX 3000 series have been much easier to buy we have seen prices settle down in 2022. In Q3 2022 you can find RTX 3080 Ti’s for around $700 to $900 and RTX 3080 for roughly $500-$600. So now that Nvidia knows what people are willing to spend they are pricing their GPUs accordingly to the market (inflation, recession etc.). The RTX 4090 will cost $1,599, the RTX 4080 (16GB) will be $1,199 and the RTX 4080 (12GB) $899. Nvidia can price their GPUs however they please, but there is more to the picture. Similar to the RTX 3080 confusing and staggered releases consisting of multiple hardware configurations; we get the same treatment this time around, but on the release day.
More Gaming Performance, but other workloads throughput could be hindered
For most enthusiasts currently using the RTX 3080\RTX 3080 Ti the RTX 4080 12GBs is simply off the table at the $900 price point. In my opinion the memory bandwidth for any GPU higher than $700 should be 512-bit by default but I can’t always have my way, however I can live with the 384-bit memory bus for the RTX 3090 and 320-bit for the RTX 3080. With the RTX 4000 series things will be changing dramatically. Only the RTX 4090 will have the 384-bit memory bus. The RTX 4080 16GB will only have 256-bit memory bus and the RTX 4080 12GB will release with only 192-bit memory bus. To keep things simple I will be using the original RTX 3080 (MSRP $699.99) for comparisons throughout this article. I will not include the 3000 series “Ti” prices. To be honest it would be difficult to use the actual price of the RTX 3080 since it has fluctuated widly over the past two years.
Focusing on the RTX 4080 16GB for a moment
The RTX 3080 that I purchased (at MSRP in 2020) has a 320-bit memory bus which moves 760 GB/s. The RTX 4080 (16GB version) will release with a 256-bit memory bus which moves 717 GB\s. The memory bus decreases by 20% and the throughput decreases by 6%. The MSRP for the RTX 3080 was $699 and the RTX 4080 16GB MSRP be $1,199. That is a whopping 71% increase not including tax. Not many people will feel good paying an extra $500.00 for inferior specs, however the performance is what truly matters right? As you can see this isn’t added up from a technical perspective, but the PC market have changed dramatically since 2020. It appears that Nvidia doesn’t want the x80 getting anywhere near the x90 in raw performance.
RTX 4080 12GB = 192-bit Memory Bus
It gets even worse for the RTX 4080 12 GB version which will only get a 192-bit memory bus. Nvidia is obviously crippling the memory bus & bandwidth. That 192-bit memory bus width is usually reserved for the lowest tier (x60) graphic cards and last gen that would have been the RTX 3060 (MSRP $329.99). The RTX 4080 (12GB) bandwidth will only be capable of moving 504 GB\s at stock. That is a 51% difference in memory bandwidth between the RTX 4080 12GB ($899) and RTX 3080 10GB ($699) or if you are a current RTX 3080 user that would be a 34% decrease in bandwidth. Now the biggest issue for current RTX 3080 users is that the cheaper RTX 4080 10GB version will cost 28% more. Let’s be honest, similar to the RTX 3000 series the RTX 4000 series will be more expensive regardless of what price Nvidia sets at MSRP. We still have AIBs that will release their cards at higher prices over the next few months and lets not forget our lovely scalpers & bots that will come into the market. This memory bandwidth decrease will definitely effect certain applications outside of gaming. Regardless of how well the RTX 4000 series runs no one wants to pay so much more while losing so much raw theoretical performance. Remember that I can easily overclock my RTX 3080 vRAM and get 840 GB\s. So comparing the RTX 4080 12GB version bandwidth to my overclocked RTX 3080 vRAM, there is a 67% difference in bandwidth throughput so in my case that would be a 40% decrease in performance in memory bandwidth if I switch to the stock RTX 4080 12GB. That’s not to say that the RTX 4080 12GB\16GB memory won’t overclock easily as well, but I am only comparison against the information that Nvidia has released.
| RTX 4000 Series SpecsOverclock-Then-Game.com | ||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
- Prev
- Next >>