We recently checked out AMD's Ryzen 7 5700G and Ryzen 5 5600G processors featuring Vega graphics in the hope that perhaps these APUs might present a valid stop-gap solution for PC gamers holding out for better graphics card pricing.
As discussed in our review, the R7 5700G doesn't make sense at the rather high $360 asking price, and we feel AMD needs to slash about $100 off that price to even consider it. Although it packs 8 cores and 16 threads, the chip packs only half the L3 cache of fully-fledged Zen 3 parts, such as the Ryzen 7 5800X, meaning the 5700G is much slower for CPU related tasks, including gaming with a discrete GPU.
The more affordable $260 Ryzen 5 5600G is better, though we still believe it's overpriced and should be positioned closer to $200. But we could at least imagine a few scenarios where the 5600G would be a viable option. Now, what many budget-conscious PC gamers were hoping for was a repeat of the Ryzen 3 3300X, but with integrated graphics for around $160. Basically a Ryzen 3 3400G replacement, but sadly we never got it, sort of.
The Ryzen 3 5300G is a 4-core / 8-thread processor featuring 6 CU integrated Vega graphics... the problem is, you can't buy it, or at least not without all the bits that make it work. You see, the 5300G is an OEM-only part, but unlike the 5600G and 5700G which spent their first 4 months of existence in OEM only status before being released into the retail channel, the 5300G is set to be an OEM exclusive indefinitely.
The reason for this likely goes down to supply, as AMD wants to avoid a repeat of the Ryzen 3 3100 and 3300X. In order to make a 5300G, AMD is using defective silicon that couldn't be binned for use as a 5600G. The Ryzen 3 only requires 4 working cores, 8MB of L3 cache and 6 CUs.
For AMD, the 5300G serves two main purposes: it allows them to turn some of that defective silicon into cash, while also allowing them to service the OEM market with budget-friendly Ryzen parts that don't require a discrete graphics card. Thus, they're not interested in releasing a retail 5300G that to meet the inevitable demand, would have to dedicate working silicon that could be used for the 5600G and possibly even the 5700G, silicon they could make them a lot more money.
This is why the Ryzen 3 3100 and 3300X disappeared from shelves so quickly. Yields were good enough that working silicon was destined for use as a Ryzen 5 3600 and Ryzen 7 3700X instead than ensuring supply of cheaper Ryzen 3 parts.
The Ryzen 3 5300G isn't a cut down version of the original Zen 3 parts, but rather it's another monolithic die, so it's a Cezanne part and is therefore a cut down version of the 5600G and 5700G, with only a quarter of the cache of the mainstream Ryzen 5000 CPUs.
With that big drop in L3 cache, no doubt there's going to be a negative impact on performance, though as we recently discovered, with only 4 cores active, you can get away with a smaller L3 cache as there simply aren't enough cores to take advantage of larger buffers. As for clock specifications, the 5300G boosts to 4.2 GHz with a base frequency of 4 GHz and that means the max boost is 200 MHz lower than the 5600G and 400 MHz lower than the 5700G.
We know why there won't be a retail version of the Ryzen 3 5300G, but if you have the chance to purchase an OEM chip, or are looking at a system using it, how well does it perform? To find out we need to get into the benchmarks, and for this we've split the testing up into three sections...
First, we're going to look exclusively at CPU performance, testing the 5300G in applications using an RTX 2080 Ti so we can compare it to the rest of our CPU data. Second, we want to see how well it works with the Vega iGPU, comparing it with a number of other integrated graphics solutions and entry-level graphics cards.
Then finally, we've also run many discrete GPU tests using the GeForce RTX 3090, allowing us to compare the 5300G to a range of other CPUs with a powerful graphics card, showing us what kind of headroom this APU offers gamers, should they upgrade down the track. Let's get into it.
Benchmarks
Starting with Cinebench, we see that the 5300G's multi-core performance is roughly equivalent to that of the 6c/12t Ryzen 5 2600, so not bad for a quad-core processor. It's also not a great deal slower than the Core i5-10400F, trailing by a 15% margin.
While the performance isn't amazing, if this were a $160 APU, I think this result would be considered very good.
The reason for the stronger than expected multi-core result when compared to parts like the R5 2600 and i5-10400F is due to the strong single core performance.
The 5300G scored 522 points and that means for single core performance in this workload it's just 4% slower than the Core i5-11400F, so a mighty impressive result there.
The 7-zip compression performance is weaker than what we saw in Cinebench, here the 5300G trailed the R5 2600 by a 10% margin and the i5-10400F by a 19% margin. That said, it was a massive 85% faster than the Ryzen 5 3400G... quite the extreme upgrade.
The 5300G falls further behind the R5 2600 for decompression work, trailing by 18%, though it was just 9% slower than the i5-10400F. Then we see that the margin over the 3400G has been significantly reduced, but even so the 5300G was a good bit faster offering 29% greater performance.
The Ryzen 3 5300G is able to lean on its strong single core performance in the Adobe After Effects benchmark, matching what are typically much faster CPUs such as the Ryzen 7 3700X and Core i7-10700K. When compared to the 3400G the new 5300G was 26% faster, so another strong performance uplift there.
We're seeing a similar thing with Photoshop which only heavily utilizes 1 - 2 cores, and this plays to the 5300G's advantage allowing it to match what are typically much faster CPUs such a the 3700X, 10600K and R5 3600.
Adobe Premiere Pro tends to use a lot more of the CPU and as a result the 5300G is located right towards the bottom of our graph with a score of just 561 points. Surprisingly, this meant the 5300G wasn't any faster than the 3400G in this test.
In Blender, the 5300G was 22% faster than the 3400G, 16% slower than the R5 2600 and almost 40% slower than its bigger brother, the Ryzen 5 5600G.
Power Consumption
An advantage of the 5300G when compared to the more powerful 5600G is that it sips power. Here we're looking at total system consumption of just 111 watts when using an RTX 2080 Ti, basically the same level of power as the 3400G, but with 20+% better performance in this test.
Integrated Graphics Performance
Moving on to iGPU testing, we'll start with Assassin's Creed Valhalla using the lowest possible quality settings at 1080p. Here the stock 5300G was good for just 26 fps on average, which is basically the same level of performance delivered by the old 3400G.
With the iGPU overclocked to 2.4 GHz, we were able to match the stock 5600G with 31 fps, a decent improvement.
The 5300G was slightly faster than the 3400G in Rainbow Six Siege, rendering 52 fps on average, an 11% improvement. Overclocking boosted performance by a further 17%, allowing for 61 fps on average, and that's slightly better than a stock 5600G.
Horizon Zero Dawn is a struggle at 1080p and for playable performance you'd probably have to drop the resolution as low as 720p. Again, overclocking boosted performance by 17% and although not terribly useful at 1080p, at lower resolutions this gain could come in very handy.
The faster clocks result in a big 32% performance uplift in Shadow of the Tomb Raider and that makes a real difference to how well the game plays, going from a very laggy 28 fps on average to a much smoother 37 fps average. Impressively, this saw the 5300G go from only matching the 3400G stock to mimicking the performance of the stock 5700G.
Moving on, we have Doom Eternal and here the 5300G was good for 40 fps, which is a solid out of the box experience. Once overclocked I was able to push it up to 47 fps, so an 18% improvement there.
Watch Dogs: Legion is another game like Assassin's Creed Valhalla or Horizon Zero Dawn that you'll need to play at 720p to avoid a slideshow. Stock the 5300G was good for just 23 fps and overclocked to the max it only provided 28 fps.
As you'd expect Counter-Strike: Global Offensive plays very well using the medium quality preset at 1080p and despite offering just 7% greater performance over the 3400G, the 5300G was good for 89 fps, so a great gaming experience.
It is possible to achieve playable performance in F1 2020 out of the box with the 5300G, rendering 46 fps on average. It's also possible to boost performance by a further 20%, pumping out 55 fps and at this point the 5300G is enabling quite an enjoyable gaming experience.
For the last iGPU benchmark we have Dota 2 and this game runs quite well on the 5300G, though we're only talking about an 11% boost over the 3400G. Meanwhile, our overclock improved performance by a further 15% to 70fps, about what the RX 550 delivers in this title.
Gaming Performance (dGPU)
Now here's how the Ryzen 3 5300G performs when paired with the GeForce RTX 3090. Granted, no one is aiming to couple those two together, but the idea is to show CPU limited gaming performance, so if you were to buy a more powerful graphics card in a few years time, this is how the 5300G would perform relative to other parts under CPU limited gaming conditions.
The F1 2020 performance certainly isn't bad with roughly 200 fps on average and a 1% low of 160 fps, almost twice that of the 3400G, so a massive upgrade there. The 5300G also beat the R5 2600 quite convincingly and wasn't a great deal slower than the R5 3600, so I'd say overall quite a good result here.
Next up we have Rainbow Six Siege, where the 5300G just edged ahead of the R5 2600 and that made it one of the slowest CPUs tested, and by some margin. Although it was a massive 42% faster than the 3400G, it was also 27% slower than the 5600G and 18% slower than the R5 3600. Overall, plenty of frames were being pushed out here despite the fact that the 5300G was much slower than modern 6c/12t parts.
Moving on to Horizon Zero Dawn which isn't a very CPU demanding title, at least by today's standards, we see that the 5300G is able to match the R5 3600, allowing it to outpace the 3400G by a whopping 36% margin. It was also just 12% slower than the Ryzen 5 5600G, so one of the better results for the quad-core 5300G.
THe 5300G appears to do well in Borderlands 3 when looking at the average frame rate, however if we turn our attention to the 1% low performance you'll see that it's significantly slower than the R5 2600 and really everything but the 3400G.
We're only talking about a 13% improvement in 1% low performance from the 3400G, making the 5300G 35% slower than the 5600G.
Performance in Watch Dogs: Legion is more of the same, the 5300G is only able to match the R5 2600 and while that meant it was still 50% faster than the 3400G when comparing 1% low performance, it was also 27% slower than the 5600G.
The Ryzen 5 3400G completely tanked in Death Stranding, dropping down as low as 41 fps for the 1% low result. That meant the 5300G was over 2.5x faster, again matching the Ryzen 5 2600, though it was a little over 20% slower than the 5600G.
The 5300G struggled relative to the 6c/12t processors in Shadow of the Tomb Raider, trailing the R5 2600 by a 9% margin and only beating the 3400G by 12%. Then when compared to the Ryzen 5 5600G we're looking at a 30% drop in performance.
In Hitman 2, the 5300G was able to edge just ahead of the R5 2600, allowing for 104 fps on average, making it 15% slower than the 5600G. Still, when compared to the old 3400G we're looking at a massive 65% increase in performance.
Here's a look at the average performance seen across the 10 games tested. As you can see, the 5300G is similar to the Ryzen 5 2600, making it 12% slower than the R5 3600, and almost 20% slower than the 5600G. Still, when compared to AMD's last retail quad-core APU, we're looking at a 45% performance improvement on average.
What We Learned
The Ryzen 5 5300G won't blow your socks off, but that was never the intention nor expectation. Compared to the Ryzen 5 3400G, which is based on the 2nd-gen Zen+ architecture, the 5300G is a huge upgrade on the CPU side, thanks to faster Zen 3 cores.
For gamers, the 5300G is a nice entry-level CPU despite only offering 4 cores with SMT support, performance was generally good and significantly better than older quad-cores such as the 3400G. There were still some instances where frame time performance wasn't great, but you're likely going to be very willing to live with that at the right price.
It's hard to award the 5300G a value rating given it's an OEM-only product, but if it were available at retail for between $150 - $170 – the same price range as the 3400G and ~$100 less than the 5600G – then it'd be a decent deal. Granted, the Core i3-10100F can often be had for just $110, but it's a slower CPU and the iGPU is basically useless for gaming, making the 5300G a more well-rounded product.
Unfortunately, we'll never get the opportunity to get excited over the 5300G as you'll have to purchase it in an entry-level OEM machine and those typically aren't very good. The 2021 HP Pavilion that I purchased to grab the 5300G is the same desktop PC we bought to score an early review of the 5700G, and technically speaking, it's not great.
Besides the fact that you have to deal with proprietary motherboard power connectors and a custom PSU, the components aren't very good. The motherboard features a weak 5-phase VRM without any cooling, so expect a limited upgrade path on that board. If you could buy the 5300G and slot it on an entry-level B550 motherboard that would make for a great combo with a decent upgrade path for either a more powerful CPU or graphics card in the future.