Wednesday, April 24th 2024
AMD's RDNA 4 GPUs Could Stick with 18 Gbps GDDR6 Memory
Today, we have the latest round of leaks that suggest that AMD's upcoming RDNA 4 graphics cards, codenamed the "RX 8000-series," might continue to rely on GDDR6 memory modules. According to Kepler on X, the next-generation GPUs from AMD are expected to feature 18 Gbps GDDR6 memory, marking the fourth consecutive RDNA architecture to employ this memory standard. While GDDR6 may not offer the same bandwidth capabilities as the newer GDDR7 standard, this decision does not necessarily imply that RDNA 4 GPUs will be slow performers. AMD's choice to stick with GDDR6 is likely driven by factors such as meeting specific memory bandwidth requirements and cost optimization for PCB designs. However, if the rumor of 18 Gbps GDDR6 memory proves accurate, it would represent a slight step back from the 18-20 Gbps GDDR6 memory used in AMD's current RDNA 3 offerings, such as the RX 7900 XT and RX 7900 XTX GPUs.
AMD's first generation RDNA used GDDR6 with 12-14 Gbps speeds, RDNA 2 came with GDDR6 at 14-18 Gbps, and the current RDNA 3 used 18-20 Gbps GDDR6. Without an increment in memory generation, speeds should stay the same at 18 Gbps. However, it is crucial to remember that leaks should be treated with skepticism, as AMD's final memory choices for RDNA 4 could change before the official launch. The decision to use GDDR6 versus GDDR7 could have significant implications in the upcoming battle between AMD, NVIDIA, and Intel's next-generation GPU architectures. If AMD indeed opts for GDDR6 while NVIDIA pivots to GDDR7 for its "Blackwell" GPUs, it could create a disparity in memory bandwidth performance between the competing products. All three major GPU manufacturers—AMD, NVIDIA, and Intel with its "Battlemage" architecture—are expected to unveil their next-generation offerings in the fall of this year. As we approach these highly anticipated releases, more concrete details on specifications and performance capabilities will emerge, providing a clearer picture of the competitive landscape.
Sources:
@Kepler_L2 (on X), via Tom's Hardware
AMD's first generation RDNA used GDDR6 with 12-14 Gbps speeds, RDNA 2 came with GDDR6 at 14-18 Gbps, and the current RDNA 3 used 18-20 Gbps GDDR6. Without an increment in memory generation, speeds should stay the same at 18 Gbps. However, it is crucial to remember that leaks should be treated with skepticism, as AMD's final memory choices for RDNA 4 could change before the official launch. The decision to use GDDR6 versus GDDR7 could have significant implications in the upcoming battle between AMD, NVIDIA, and Intel's next-generation GPU architectures. If AMD indeed opts for GDDR6 while NVIDIA pivots to GDDR7 for its "Blackwell" GPUs, it could create a disparity in memory bandwidth performance between the competing products. All three major GPU manufacturers—AMD, NVIDIA, and Intel with its "Battlemage" architecture—are expected to unveil their next-generation offerings in the fall of this year. As we approach these highly anticipated releases, more concrete details on specifications and performance capabilities will emerge, providing a clearer picture of the competitive landscape.
114 Comments on AMD's RDNA 4 GPUs Could Stick with 18 Gbps GDDR6 Memory
There will be future videos that compare 20GB RX 7900 XT vs 16GB RTX 5080, and guess what, the results will be shocking, but against the black-leather-jacketed guy.
Stay information that a RTX 5080s only have 16 VRAM....Not good deal with GDDR7...
We need a new mid-range focused generation to raise the baseline GPU performance so that many more people will be able to play next gen titles at playable framerates at decent settings, and may even dabble in RT in lighter titles/at lighter settings.AFAIK, These are the most likely specs for N48 and N44
WGPs
CUs
Memory Bus
??? (Memory Bandwidth in GB/s?)
??? (Base Clocks? N44 has an error, assume it's 2515)
Die Size
18 Gbps over 128-bit is even worse - only 288 GB/s memory throughput.
Maybe they want to start with a weak line-up again, and a year later they're going to do a Super refresh with 3 GB modules (so a 24 GB 5080 S and 18 GB 5070 S).
But if RDNA4 tops out at ~7900 XT performance, it's not a good upgrade either.
Can you imagine the fail that the RX 7700S would have been if AMD crippled it with only 8GB?
*checks Adrenalin*
All of that depends on the die size and what sort of cost the memory will be. Looking at SK hynix financials recently was a bit of a wake up call!
And that's the whole point. No GPU vendors are able to penetrate the market, if they will follow the nVidia's flute. They must disregard nVidia, as they disregard everyone else. AMD and Intel must push their own game, despite competition. They must do their best, as there's no rivals are. This includes undercutting nVidia, and follow their own budgets, and their BOM and requirements, and set the pricing accordingly. Not like now, where AMD has the tech like much cheaper to R&D and manufacture, but somehow manages to sell the products with the pricetags of nVidia's R&D budget and premium. This doesn't work. AMD is in the situation, when they must sell their cards cheaper, and sacrifice margins, in order to get the marketshare. This is just stupid. AMD just teased people with their decent cards like 6800XT/7900XT, and they just drop the advancement, in hi-end, and even fully competitive mid-end, first time for the decade, and bail out so abruptly. And the sales, availability and marketing is stagnant and uncertain to say the least. AMD has to use the situation, while people are warmed for AMD products, and make some refresh of monolithic RX6800XT/6900XT, with some architecture improvements, and maybe on smaller more efficient node, and sell it as e.g 8600XT/8700XT for about $300 and $350-400, and the sales would be huge.
Intel priced their card below the competition. Yes their current products are lacking, and are inferior at power efficiency and high performance. But if they will be able to keep the smae full steam pace, and deliver the better products on the round three, they have a chance to gain much bigger mindshare, marketshare, and momentum and turn the tables drastically, and swiftly. This might be wishful thinking, but IMHO, Intel, has more theoretical chances to overcome AMD, with same or bigger marketshare, than AMD with their sluggish marketing and management. Considering how they fix their drivers, and even do complete overhaul, after overhaul, being first time into dGPU market...
Don't get me wrong, though, AMD has really nice products. But they have neither capacity, nor advantage, speed, and most importantly the wish to shake the market. Their products are either missing, not available for purchase, or their price is awful. AMD have not control over the partners, which can screw up them up, the damage contro and quality contro are non-existant. It just looks, like they don't even care about consumer GPU division at all.
And that's why Intel has more chance. Even without dirty tricks and bribes, and own foundries, Intel has stronger corporate stucture, are more known and widespread and OEMs will gladly jump their ship, abandoning AMD, as soon, as Intel will make a breakthrough.
Even if AMD will keep medium-range-only GPU products, they have to deliver in time, have the agressive pricing and be on par, or even have slight advantage. But tha'ts hard, and maybe that's why they don't even try anymore with Radeon division. After all, the APU OEM sales with AI give them certain high marging flow of money.
It depends on you - upgrade your monitors.
AMD has cards that go all the way up to a bit above 4080 level performance, a performance level of which no one finds embarrassing. Far from it, both the 4080 and 7900 XTX are extremely capable cards. The vast majority of people cannot even afford a card of that level of performance.
You seem to be under the influence of the halo effect, that because AMD doesn't have the performance leading card at the top of the stack somehow AMD's entire lineup is bad.
There's certainly an argument to be made to favor Nvidia for it's features but to argue that AMD's performance is embarressing ignores the fact that AMD matches Nvidia performance through it's entire stack excluding only the 4090. Unless you declare that every card slower than a 4090 is embarssing performance wise (which is a ridiculous claim), you have an extremely obvious double standard.
Most gamers are in fact not on the ray tracing bandwagon. Both HWUB and GamersNexus did a poll and less than 30% of enthusiasts consider ray tracing an essential factor when purchasing a video card. That's just among enthusiasts as well, the actual rate at which ordinary people care is without a doubt even lower (especially given the average PC gamer doesn't even have a card that can use RT at all or can't do so at an acceptable FPS given the cost of well performing RT capable cards). The idea that everyone is on the RT bandwagon is completely incorrect.
I'd also add that even if his argument was that AMD is embarrassing performance wise due to it's RT performance (which again wasn't his original argument and isn't even mentioned in his post. It's just something you fabricated to try and defend his comment for some unknown reason), the 7900 XTX is only 17% slower in RT performance. That's not what I call embarrassing, not by a long-shot.
Neither the original argument or the goal post shifted argument 'he must be referring to RT performance' have any ground to stand on. Utter tripe. I don't see the point in jumping in the line of fire for an argument that was almost certainly made with disingenuous intent.