Intel Arc Alchemist Release Date, Price, Performance, Specs

In the midst of two GPU juggernauts hyping up their new releases, a less prominent name in the field is attempting to enter the fray. Intel hopes to finally be competitive in the GPU market with its Intel Arc Alchemist lineup.

The rivalry between AMD and NVIDIA is heating up again after years of NVIDIA dominance. However, things appear to be on far more equal footing following the releases of their next-gen graphics cards.

All of this provides a perfect storm for Intel to enter the market.

Let’s get started!

Related:AMD vs. NVIDIA – Which GPU Manufacturer Should You Choose?

Release Date

Originally, Intel was supposed to release the Arc Alchemist GPUs quite a bit sooner, but because of numerous GPU or management-related issues, the graphics cards were delayed numerous times.

Fortunately, the Intel Arc A750 and Arc A770 were finally released on October 12th, 2022.

The rest of the lineup is still in the dark, with the exception of the Arc A380, which was released on August 16th, 2022.

Supposedly, Intel planned to release an Alchemist refresh in 2023, however, a known leaker, MLID, recently shared rumors that this refresh might be canceled.

That is unfortunate considering that a refresh might’ve put Intel in an even more competitive situation in the GPU market.

But, this is just a rumor, so there is still a possibility that Intel does fulfill its original Alchemist refresh plan.

If you are not interested in this generation of Intel Arc Alchemist GPUs, hold out for future generations of Intel Arc like Battlemage, Celestial, or Druid.

Price

Since the original announcement of Intel’s desktop GPUs, Intel’s idea was to offer the most competitive pricing on the market, and it seems they succeeded.

Here’s the price list, but do note that certain GPU MSRPs are subject to change once they release:

• A310 – <$100?
• A380 – $139
• A580 – $220
• A750 – $249 ($289 MSRP)
• A770 8GB – $329
• A770 16GB – $349

With the 4086 driver update, Intel also announced that the price of the A750 is officially reduced to $249 from the $289 MSRP.

Overall, Intel ARC GPUs offer great value but still lack in performance in certain video games (and features), but Intel is continously working on improving their drivers.

For example, with Intel’s 4311 driver, the Arc A750 offers considerably better perforamnce per dollar than the RTX 3060 in Total War: Warhammer III – Mirror of Madness.

Performance

With the initial release of the low-end Arc A380, the future for Intel’s Arc Alchemist lineup looked pretty grim.

But, the release of the A770 and A750 showed that Intel’s GPUs indeed offered solid performance for a good price.

However, it’s not all sunshine and rainbows for ARC Alchemist.

Driver Issues/Updates

After the launch of the Intel Arc A750 and A770 GPUs, various reviewers and users reported poor driver performance.

This led to crashes, failure to run certain games, monitor detection issues, abysmal performance, especially in DX9 titles, and more.

However, Intel is working diligently to improve Arc drivers as best as possible. Recently, the v3959 driver update significantly improved DirectX9 performance.

The driver also brought huge improvements to the minimum FPS and frametimes, so the overall experience for Arc users should be much smoother.

Intel’s February 2023 driver update also brought some massive gains in FPS, especially in DX9 titles.

Intel claims that the Arc A750 is 77% faster in Counter-Strike: Global Offensive with the new 4086 patch. These are some huge improvements and it shows that Intel is 100% invested in its graphics market.

In this DigitalTrends interview, Tom Peterson and Omar Faiz cover the topic of how Intel’s GPU drivers have improved so much over such little time.

Intel’s version 4311 driver further improves the A750. In Dead Space (2023), the A750 with the latest driver is 62% faster than the previous version 4257 driver!

Furthermore, the new driver makes the A750 faster than the RTX 3060.

In short, Intel originally used a DX12 to DX9 translation layer (D3D9on12), which proved as a subpar solution and the only way to resolve that was to start working on DX9 support from zero.

Here’s an in-depth before-after performance comparison from Hardware Unboxed:

High-End Desktop SKU

While the low-end A380 has provided disappointing results, Intel’s Arc high-end shows promise.

A770

To start on a good note, here’s a Gamers Nexus chart of Strange Brigade (in Vulkan) in which the Arc A770 performs quite well against its competitors.

The 16GB A770 is about 25% faster than the RTX 3060 and about 7% slower than the RTX 3070. For a $349 card, those are some impressive numbers.

If the Intel ARC lineup performs as well in every game as in Strange Brigade, this entire article would be a different story. Unfortunately, that is not always the case.

Here are F1 2022 and Tomb Raider as examples:

Here, the Arc A770 is considerably slower than the AMD RX 6600, a card that’s available for around $250. Already that’s a huge disadvantage for Intel.

However, Intel’s hard work on the Arc drivers has improved its performance in certain games, especially in competitive titles.

Here’s the A770 with 2 different patches in CS: GO.

With the 4091 patches, the A770 offers more than double the FPS compared to A770 with the release patch.

So, the $350 16GB Arc A770 now looks like a much better option, considering that it now performs better than the RTX 3060 in various games.

A750

In most video games, it seems the A770 is only about 5% faster than the cheaper A750. So, is it really worth paying 20% extra for such a small increase in performance and plus 8GB of VRAM?

In Watch Dogs: Legion, we see some good results for the A750. It outperforms the RTX 3060 and nears the RTX 3060 Ti, but is still quite a bit behind the RX 6600 XT.

However, by increasing the resolution to 1440p, both Arc GPUs get a bump in performance.

With the latest driver, the Intel Arc A770, in Strange Brigade (a Vulkan API title) is only about 10% slower than the RTX 3070.

Mid-Range GPU – A580

The A580 will be Intel’s mid-range and possibly the best price-per-performance GPU.

We barely have any information for this SKU, but recently an Ashes Of The Singularity benchmark leaked, so we can use that as insight.

According to VideoCardz, the RTX 3060 scores anywhere between 75 to 96 FPS in this same (Vulkan) benchmark, so performance with the A580 is pretty similar.

Keep in mind, though, the A580 isn’t an RTX 3060 competitor, so take this leak with a bit of salt.

Low-End – A380

The A380 is the first desktop GPU to release from the Intel ARC lineup, which is a complete letdown in almost all aspects.

Here’s why:

The RX 6400 comes out on top by more than 10%. It’s not a huge difference in performance, but when you consider that this is a brand new architecture and a price point of ~$140.

Currently, Intel’s low-end GPUs are not very appealing, especially after you consider the awful performance in older titles (CS: GO, Rainbow Six Siege, etc.)

Intel’s Three-Tier Gaming Strategy

The reason why we’ve seen such bad performance in games that use older APIs like DX9 and DX11 is that Intel has a “Three Tier Game Strategy” in place.

They will be splitting games into three different tiers, and here’s how that split will look:

• Tier I: Games that have support for the latest APIs and are optimized for Intel’s Arc architecture. So, newer games like Cyberpunk 2077, Tomb Raider, Call of Duty, etc. This is where Intel will offer the best price/performance.
• Tier II: Games with DirectX12 and Vulkan, but without optimizations for Intel Arc GPUs.
• Tier III: Videos games on older APIs will run worse on Intel Arc GPUs compared to AMD’s and Nvidia’s graphics cards in the same price range.

There are definitely bad and good sides to this. If you consistently play the latest titles, you will most likely have great performance when compared to AMD or NVIDIA.

However, if you are a person that enjoys older titles, understand that an Intel Arc GPU may not perform as well as you would like.

Specifications And Features

With the Intel Arc A770, A750 and A380 released, we already have a good idea of their performance.

But, let’s look at some specifications and, more importantly, unique features.

SKU/GPU Name	Xe Cores	Shading Units	Clock Speed	Memory	Memory Bus	TBP
Arc A770 16GB	32	4096	2100 MHz	16GB GDDR6	256-bit	225W
Arc A770 8GB	32	4096	2100 MHz	8GB GDDR6	256-bit	225W
Arc A750	28	3072	2050 MHz	8GB GDDR6	256-bit	225W
Arc A580	16?	2048?	?	8GB? GDDR6	128-bit?	~150-200W?
Arc A380	8	1024	2000MHz	6GB GDDR6	96-bit	75W
Arc A310	6	?	2000MHz	4GB GDDR6	64-bit	75W

Currently, the A550 still has not been announced or any other Arc 5-series GPU. The A310 was announced, but still no exact release date.

Resizable BAR

Resizable BAR (Intel and Nvidia) or Smart Access Memory (Ryzen+Radeon) has been a thing for a few years now. It’s a reintroduced feature that allows for better communication between the CPU and the GPU.

Weirdly enough, in Intel’s Arc quick start guide for desktop SKUs, Intel specifies that Resizable BAR is a must-have for the best possible performance out of this new generation of GPUs.

However, ReBAR is available only on Intel’s 10th, 11th, and 12th generation CPUs. Does this mean that Intel Arc GPUs won’t perform as well on older Intel platforms or on AMD CPUs?

This is yet to be tested by reviewers.

Xe-Cores

Intel introduced us to its new desktop GPU architecture and goals with Intel’s Architecture Day 2021.

After an almost never-ending silence (or more) on Intel Arc, we got information on Arc alchemist. Specifically, Intel’s compute building block, Xe-Core. The Xe-Core is a direct replacement for Intel’s Execution units (EUs).

Every Xe-core includes 16 Vector Engines (VE) and 16 Matrix Engines (XMX). Each of those Vector Engines processes 256 bits per cycle. Based on Intel’s explanation, each VE has 8 ALUs. That leads to a total of 128 ALUs in one Xe-Core. The core also has its L1 cache.

Intel combines four of these cores to create a Render Slice. We’ve now jumped up to 512 ALUs and 64 XMX.

Furthermore, there are four Ray Tracing Units in each Render Slice to help with ray tracing performance. There are also Samplers, Pixel Backends, and engines for Geometry, Rasterization, and HiZ.

However, Intel had to scale it further to see the true potential of these cores and Render Slices. Thus their flagship GPU holds eight Render Slices. That means 32 Xe-Cores, 512 Vector, and Matrix Engines, and a total of 4096 ALUs.

Here’s a Gamers Nexus video with Intel’s Graphics Engineer Tom Petersen explaining the engineering behind the Arc GPUs:

Intel XeSS Upscaling

Those Matrix Engines we mentioned previously have a role similar to NVIDIA’s Tensor Cores which use to accelerate AI-related workloads. That is why DLSS 2.0 is so good at supersampling.

Intel’s XeSS upscaling can utilize XMX to provide similar upscaling to DLSS, but the difference is that it also works with NVIDIA and AMD GPUs too.

However, XeSS didn’t start out great. But, we have to keep in mind that DLSS and FSR didn’t perform too well until they got updated.

With future updates, we hope that Intel will be able to bring XeSS’ quality and performance to NVIDIA and AMD’s levels.

Here are some charts so you can see how this new technology performs:

With an RTX 3060 at 1440p, XeSS performs about 17% worse than DLSS 2.4 and FSR 2.1 at the Ultra Quality preset.

At 4K with an AMD card (RX 6650 XT), the performance gap between XeSS and FSR becomes even larger. In Spider-Man Remastered, with the Ultra Quality/Quality preset, XeSS is almost 40% slower than FSR 2.1.

That’s not all though as visual quality is also affected by XeSS, so Intel graphics team definitely still has some work to do.

If you want to see the difference in visual quality between these three technologies, here’s Hardware Unboxed’s video:

It’s not all bad though!

Hardware Unboxed did some more testing on XeSS and came to the conclusion that this technology works better with Arc GPUs (as expected.)

Visual quality and performance boost is better when compared to the previous tests with an RTX 3060 and RX 6650 XT.

Take a look:

In Hitman 3, with an Arc A770 at 4K, the XeSS Quality preset delivers about 8% more performance without any significant loss in image quality.

If the resolution is dropped down to 1440p, XeSS still holds the lead, but only by a very small margin.

If you would like to properly see a visual comparison, check out their video:

Intel Deep Link, AV1 Encoding, And Arc Control

Another interesting new feature from Intel is the full support for AV1 encoding and decoding. That’s something that neither Nvidia nor AMD have, so that’s a big plus for content creators.

Combine the speed of AV1 encoding (already faster than standard HEVC encoding) with Intel’s new Deep Link technology that helps the CPU and GPU share power/resources, which leads to even faster encoding performance.

AV1 encoding might be one of the only good sides of Intel Arc GPUs, mostly because it is open source and performs better than NVIDIA’s NVENC and H.264 encoders.

EposVox uses Netflix’s VMAF, a video quality benchmark, to determine which video coding delivers the best possible viewing experience.

Looking at this chart, it is pretty obvious that Intel’s AV1 encoder is miles better than Nvidia or AMD. At 3500 kbps bitrate, it has an 83 VMAF score compared to NVENC, which scored only 71.

At a higher bitrate, the gap in video quality between these encoders gets smaller, but Intel’s AV1 is still the best out of the bunch.

Either way, streamers can use a low 3.5 MB/s bitrate and still offer video quality comparable to NVENC at 8.0 MB/s

Paired with these new features will be Intel’s new Arc Control software which can be used for recording, updating drivers, tweaking game settings, and more. It will have a role similar to GeForce Experience and AMD’s Radeon Software.

Intel Arc Alchemist Mobile

While most of this guide focuses on Intel’s desktop releases because that’s what most people have been waiting for, we can’t ignore the fact that Alchemist will show up on laptops too.

Here’s a table for a bit more specifications.

	A350M	A370M	A550M	A730M	A770M
Xe-cores	6	8	16	24	32
Ray Tracing Units	6	8	16	24	32
GPU Clock	1150 MHz	1550 MHz	900 MHz	1100 MHz	1650 MHz
GDDR6 Memory	4GB	4GB	8GB	12GB	16GB
Memory Bus Width	64-bit	64-bit	128-bit	192-bit	256-bit
GPU Power	25-35W	35-50W	60-80W	80-120W	120-150W

VideoCardz’s original leaks have an exact match with these specifications shared on Intel’s Arc Event.

So, the A770M, the flagship GPU, gets 16GB of GDDR6 VRAM and a power of up to 150W, allowing a GPU clock of 1650MHz. The weaker A730M clocks quite a bit lower and gets 12GB VRAM.

The mid-range A550M comes with 8GB VRAM with a 1600MHz boost clock.

The little brothers, the A350M and A370M, both come with 4GB of VRAM and clocked at 1150MHz and 1550MHz, respectively.

We might not be witnessing a full-on three-way war between Intel, AMD, and NVIDIA, but something exciting is brewing. As gamers and tech enthusiasts, we can’t wait to see it.

Mobile GPU Performance

Both the A730M and A770M come in Alder Lake CPUs, while the RTX 3050 Ti and RTX 3060 are paired with Intel’s 11th-Gen CPUs.

So, here’s the comparison, but make sure you take it with a bit of salt:

Games	RTX 3050 Ti	Arc A730M	RTX 3060	Arc A770M	A730M vs 3050 Ti	A770M vs 3060
Assassin's Creed Valhalla @ High	38	50	74	69	132%	93%
Borderlands 3 (Ultra)	45	50	60	76	111%	127%
Control (High)	42	62	70	89	148%	127%
Cyberpunk 2077 (Ultra)	39	49	54	68	126%	126%
Death Stranding (Ultra)	89	87	113	102	98%	90%
Dirt 5 (High)	64	61	83	87	95%	105%
F1 2021 (Ultra)	68	86	96	123	126%	128%
Far Cry 6 (Ultra)	63	68	80	82	108%	103%
Gears of War 5 (Ultra)	58	62	72	73	90%	101%
Horizon Zero Dawn (Ultimate Quality)	63	50	80	68	79%	85%
Metro Exodus (Ultra)	39	54	53	69	138%	130%
Red Dead Redemption 2 (High)	46	60	66	77	130%	117%
Strange Brigade (Ultra)	98	123	134	172	126%	128%
The Division 2 (Ultra)	63	51	78	86	81%	110%
The Witcher 3 (Ultra)	96	101	124	141	105%	114%
Total War Saga: Troy (Ultra)	48	66	71	86	138%	121%
Watch Dogs Legion (High)	59	71	77	89	120%	116%
17-Game Geometric Mean	57,2	64,6	78,8	88,3	113%	112%

This is a solid variety of games, including both Nvidia and AMD optimized titles.

Based on Intel, the A730M outperforms the RTX 3050 Ti in almost all games, resulting in 13% more FPS. Notable gains over the 3050 Ti are in Metro Exodus, Control, and Total War Saga: Troy.

The A770M is supposedly 12% faster than the notebook RTX 3060. Overall, this entire table of numbers favors Intel’s Arc lineup, but can we really trust it? Also, we have to consider the cost of the laptops to get the full picture.

The recent leaks from 3dMark match these benchmarks.

The A770M scored 13244 points in Fire Strike Extreme, while the mobile RTX 3070 scored around 14000.

But is that really enough? Is all of this hype just for Intel Arc’s flagship to end up slower than Nvidia’s mid-end RTX 3070?

Here is also the lower mid-end Arc A550M in a Time Spy benchmark:

To put this into perspective, the RTX 3060 (notebook version) puts between 9000 and 10000 graphics score. It’s even slower than the RTX 3050.

For any future updates on Intel’s Arc desktop or mobile GPUs, make sure to check back on this article!